EIOPA-TFQIS5-11/001

14 March 2011

EIOPA Report on the fifth

Quantitative Impact Study (QIS5)
for Solvency II

EIOPA – Westhafen Tower, Westhafenplatz 1 - 60327 Frankfurt – Germany – Tel. + 49 69-951119-20

Fax. + 49 69-951119-19, Website: https://eiopa.europa.eu
© EIOPA 2011
Table of contents
Executive Summary ............................................................................................................. 4
1. Introduction ................................................................................................................ 19
1.1. Disclaimer............................................................................................................. 19
1.2. Background .......................................................................................................... 19
1.3. Objectives............................................................................................................. 20
1.4. Participation.......................................................................................................... 21
2. Overall financial impact ............................................................................................. 23
2.1. Overall surplus ..................................................................................................... 23
2.2. Breakdown of the surplus by country .................................................................... 28
2.3. The main drivers of the surplus changes .............................................................. 29
2.4. Impact of diversification ........................................................................................ 31
2.5. SCR coverage ...................................................................................................... 32
2.6. MCR coverage ..................................................................................................... 33
2.7. Group surplus ....................................................................................................... 34
3. Valuation of assets and liabilities other than technical provisions ........................ 35
3.1. General ................................................................................................................ 39
3.2. Impact .................................................................................................................. 40
3.3. Materiality ............................................................................................................. 40
3.4. Mark to model ....................................................................................................... 41
3.5. Intangible assets................................................................................................... 41
3.6. Deferred Taxes ..................................................................................................... 42
3.7. Contingent liabilities .............................................................................................. 42
3.8. Financial liabilities (other than technical provisions).............................................. 42
3.9. Pension liabilities .................................................................................................. 43
3.10. Investment funds .................................................................................................. 43
4. Technical provisions.................................................................................................. 44
4.1. Comparison with current regime ........................................................................... 44
4.2. Discount rate and illiquidity premium .................................................................... 45
4.3. Risk margin .......................................................................................................... 48
4.4. Segmentation ....................................................................................................... 53
4.5. Contract boundaries ............................................................................................. 56
4.6. Other feedback ..................................................................................................... 57
4.7. Reinsurance recoverables .................................................................................... 60
5. SCR – Standard formula ............................................................................................ 63
5.1. The overall SCR ................................................................................................... 63
5.2. Single equivalent scenario methodology ............................................................... 68
5.3. Loss absorbing capacity of technical provisions and deferred taxes ..................... 69
5.4. SCR Aggregation and operational risk .................................................................. 71
5.5. Market risk ............................................................................................................ 71
5.6. Counterparty Default Risk ..................................................................................... 76
5.7. Life Underwriting risk ............................................................................................ 77
5.8. Health underwriting risk ........................................................................................ 80
5.9. Non-life Underwriting risk ...................................................................................... 86
5.10. Undertaking-specific parameters .......................................................................... 94
5.11. Risk mitigation .................................................................................................... 100
5.12. Participations ...................................................................................................... 103
6. SCR – Internal model ............................................................................................... 106
6.1. Internal models on solo level .............................................................................. 106
6.2. Current status of internal modelling in the EEA ................................................... 107
6.3. General comparison of the internal model results with the standard formula ...... 113
6.4. Partial internal models ........................................................................................ 115
7. MCR........................................................................................................................... 117
7.1. MCR calculation ................................................................................................. 117
7.2. MCR corridor ...................................................................................................... 117
7.3. AMCR................................................................................................................. 118

Page 2 of 153
© EIOPA 2011
8. Own Funds ............................................................................................................... 120
8.1. Composition of Own Funds ................................................................................ 120
8.2. Unrestricted Tier 1 .............................................................................................. 121
8.3. Comparison with Solvency I ............................................................................... 123
8.4. Other paid-in capital instruments ........................................................................ 123
8.5. Adjustments to Basic Own Funds ....................................................................... 126
8.6. Ancillary Own Funds ........................................................................................... 129
8.7. EPIFP ................................................................................................................. 131
9. Groups ...................................................................................................................... 133
9.1. Participation and methods tested........................................................................ 133
9.2. Comparisons of the various methods and principal results ................................. 134
9.3. SCR standard formula - diversification effects .................................................... 137
9.4. Availability constraints on group own funds ........................................................ 141
9.5. Other topics ........................................................................................................ 142
10. Practicability and preparedness.......................................................................... 143
10.1. Preparedness ..................................................................................................... 143
10.2. Practicability ....................................................................................................... 144
10.3. Guidance ............................................................................................................ 146
10.4. Technical Provisions simplifications .................................................................... 146
10.5. SCR simplifications ............................................................................................. 148

Page 3 of 153
© EIOPA 2011
Executive Summary

A. Background to Solvency II and the fifth quantitative impact study

Solvency II is a regulatory project that provides a risk-based, economic-based and

principle-based framework for the supervision of (re)insurance undertakings. It
acknowledges the main characteristics of the (re)insurance sector by building upon
them. In Solvency II, capital requirements will be determined on the basis of the risk
profile of undertakings, as well as on the way in which such risks are managed,
therefore providing the right incentives for sound risk management practices and
enhanced transparency.

Solvency II is a long-term project that started more than ten years ago, building a
reference regulatory framework that will apply both in normal and crisis
circumstances, from 2013 onwards. During its development relevant lessons from the
last financial crisis have also been incorporated: the fifth Quantitative Impact Study
(QIS5) takes into account a number of lessons learned from the recent financial crisis.

The design of the framework relies on technical provisions which allow undertakings to
meet their commitments towards policyholders arising from the (re)insurance activity
(i.e. the expected obligations), and capital requirements which should cover
unexpected losses over a one-year time horizon. Undertakings will have to hold
sufficient financial resources to absorb losses and to meet the risks: basic own funds
and ancillary own funds will be classified into three tiers depending on their
permanent availability and their subordination, ensuring that most resources are of
the highest quality.

A supervisory ladder of intervention is embedded in the system, by setting two target

levels of capital: the Minimum Capital Requirement (MCR) and the Solvency Capital
Requirement (SCR). Whereas the Solvency Capital Requirement incentivises sound
risk management through the explicit quantitative measurement of the risks for the
undertaking‟s operations and investments, the Minimum Capital Requirement should
ensure a supervisory response to the degradation of the undertakings‟ financial
position, allowing for ultimate supervisory action, including withdrawal of the license.
The framework is completed with the existence of a number of dampeners, both
quantitative and qualitative, that aim to address potential procyclical effects of the
regime.

Fully in line with the requirements of risk-based supervision and regulation, Solvency
II removes the implicit prudence embedded in technical provisions currently existing
in Solvency I, and provides with a fully comprehensive approach to (quantifiable) risks
within the SCR standard formula, as compared to the simplistic factor-approach taken
for the determination of the required solvency margin in Solvency I.

The starting point of the solvency assessment under Solvency II is the harmonised
solvency balance sheet valued according to market consistent principles. This
harmonised balance sheet differs from the one in the audited accounts used under
Solvency I.
The total balance sheet approach requires a consistent valuation of assets and
liabilities. This approach implies that where adjustments are being made to one of the
balance sheet items under Solvency II, this will affect the overall solvency position of
the undertaking, measured by the net asset value (assets minus liabilities).
Page 4 of 153
© EIOPA 2011
The most immediate difference is the average increase of the level of own funds as
compared to Solvency I due to a simultaneous decrease of technical provisions and,
depending on the current accounting GAAPs applied in participating countries, an
increase in the values of assets.
At European level the SCR increases when compared to the current required solvency
margin of the Solvency I system, with the Minimum Capital Requirement (MCR) in
Solvency II being below or close to the required solvency margin under Solvency I.

Furthermore, providing the right incentives for better risk management is paramount
to the system. Sound risk management is incentivised in Solvency II through for
example the possibility for undertakings to use undertaking-specific parameters
(USP), the recognition of diversification benefits and risk mitigation techniques or the
allowance, for the calculation of regulatory capital, of partial and full internal models,
subject to supervisory approval.

Due to these differences, Solvency I and Solvency II, being designed differently,
cannot be simply compared without acknowledging such differences.

QIS exercises are crucial to the development of EU regulation. QIS5 is the fifth in the
sequence and probably the last fully comprehensive exercise. The QIS exercises are
essential to strive to ensure that Solvency II is designed in the most appropriate
manner, with sufficient evidence of the impact of the regime proposed.

We note that QIS5 is a field test and not a proposal for the final Solvency II
framework. Furthermore caution should be exercised when drawing conclusions from
the figures given in this report, since the comparability of results has in some cases
been impacted by differences in the interpretation of requirements and by the short
timescales in which data had to be provided.

Solvency II will not be a perfect system the day it enters into force, yet it will be a
sound one, subject to improvements on the basis of new evidence. EIOPA, together
with the European Commission and all relevant stakeholders, keeps working in full
transparency on those areas where there is room for improvement. The findings of
QIS5 will feed into the ongoing and future work for the Level 2 Implementing
Measures that will put into practice the Solvency II Level 1 Framework Directive.

B. Participation rate

In its Call for Advice to CEIOPS, the European Commission has set out an ambitious
target participation rate of 60% for solo undertakings and 75% for insurance groups.
Once again, thanks to the continuous close cooperation of European trade
associations, long-time stakeholders and the efforts of national supervisors EIOPA has
outperformed these targets despite the tight time frame. Overall, EIOPA has
witnessed through the QIS5 participation rate an increase of the attention to the
Solvency II project.

Through five QIS exercises (and one preliminary field study) carried out in the last six
years, the number of participants has increased steadily, to a point where today 68%
of the (re)insurance undertakings that are likely to be under the Solvency II scope
have participated in this exercise. Compared to QIS4, an overall increase of the
participation of 78% can be observed.

 All 30 EEA member countries are represented in the scope of this study.
Page 5 of 153
© EIOPA 2011
 In total, 2,520 (re)insurers as well as 167 groups have participated in this study,
compared to 1,412 and 106 respectively in QIS4.

 In total, more than 95% of technical provisions and 85% of premiums of the
insurers subject to Solvency II are covered by the test.

 It was especially notable that the number of small undertakings that took part in
the study more than doubled compared to QIS4. This could also be observed in the
increased participation of medium and small-sized groups.

 In particular, the increase in the participation of reinsurers should be pointed out.

This shows that EIOPA has succeeded in engaging with supervisors and industry in the
regulatory and supervisory discussion, which will continue to benefit the
implementation efforts in the future months. It indicates that Solvency II has become
a priority to all insurers, regardless of size, and that (re)insurance undertakings and
groups are striving to be ready for the implementation date of 1st January, 2013.

C. Financial impact

Surplus

Two main elements explain the financial situation of the (re)insurance industry
resulting from the QIS5 field test at the end of 2009: the impact of the financial crisis
and the difference between Solvency I and Solvency II solvency balance sheets.

It is a fact that the financial crisis was not originated by (re)insurers and that they
have resisted much better than other financial institutions the effects of the crisis.
However, it is also a fact that, since 2007 - the basis for the previous QIS4 exercise -
the financial surplus of the insurance sector, calculated under Solvency I rules (i.e.
neutral of any Solvency II implications) has decreased markedly in 2008 (minus
€200bn), and was followed by partial recovery in 2009 - which constitutes the basis
for the current QIS5 exercise. This evolution is largely explained by the impact of the
financial crisis on the valuation of the assets owned by the sector. At the end of 2009,
the capital surplus of the insurance and reinsurance industry totalled around €500bn
compared to over €600bn at the end of 2007.

The results of QIS5 are also driven by the fundamental difference of valuation of the
balance sheet and the meaning of the solvency requirements under Solvency II as
explained above, which globally leads to an increase in capital requirements, a
decrease in technical provisions and a relative increase in the amount of eligible own
funds.

Taking into account these elements, the financial position of the European
(re)insurance sector assessed against the QIS5 solvency capital
requirements calculated in accordance with the standard formula or internal
models remains comfortable with eligible own funds in excess of the regulatory
requirements by €395bn. This amounts to a decrease of the surplus of €56bn
compared to the current regime. On a global level, the surplus under QIS5 is roughly
12% lower than the current surplus.

Page 6 of 153
© EIOPA 2011
On a national level, the evolution of the surplus is not homogeneous. In thirteen
countries the capital surplus assessed against the QIS5 SCR is greater than the
current surplus assessed against the Solvency I required solvency margin.

Generally, across all solo respondents the SCR results obtained by using an internal
model were very close to those derived by applying the standard formula. The most
significant difference between standard formula and (partial) internal model results
was observed among groups. Groups‟ internal model results showed a capital
requirement of about 0.8 times the size of the capital requirement based on the
standard formula calculation.

Compared to Solvency I, for groups using the accounting-consolidation method with

the QIS5 SCR standard formula calculation a reduction in the group surplus of around
€86bn has been observed (from €200bn to €114bn), which represents a reduction of
43% compared to the Solvency I surplus. For groups that submitted internal model
results, an increase in surplus of about €6bn has been observed when moving from
Solvency I to QIS5, which represents an increase of 6%. However, it should be noted
that there is a high variability in the results in this area.

On average, when groups applied the deduction & aggregation method, rather than
the accounting consolidation-based method, there was a significant loss of surplus.
This is due to the non-recognition of diversification effects.

For groups with entities located in non-EEA countries, results have shown a significant
impact on the group overall surplus determined by the application of local rules
instead of Solvency II rules when using the deduction aggregation method. Based on
approximations, the overall positive impact of the use of local rules for non-EEA
entities using this method is about €45bn.

Key indicators of the SCR shock and coverage of the SCR and MCR

The sum of all risks modelled under the SCR requirements calculated using the
standard formula or full or partial internal models in QIS5 totalled more than
€1300bn. Taking into account the reduction arising from diversification benefits
recognised at solo level based on the correlations between the risks (€466bn) and the
adjustment recognizing the undertakings‟ ability to reduce discretionary benefits or to
pay less taxes after a stress (in total €314bn), this leads to the final SCR being a little
above 41% of the sum of all risks modelled (€547bn).

On average, the main risk drivers of the SCR are the market sub-risks (equity, spread
and interest rates) followed by the non-life underwriting sub-risks (premium and
reserve risk and catastrophe risk).

At European level, 15% of the participants did not fully cover the SCR, which would
trigger regulatory action. Fewer than 9% of participants covered 75% or less of the
SCR. A quarter of those undertakings belong to insurance groups or financial
conglomerates for which a capital reallocation or intra-group risk transfers would be
available as a means for raising their capital level.

Just under 5% of the participants did not fully cover the MCR, which would trigger the
most serious intervention from the supervisor, this is the withdrawal of the license.

Page 7 of 153
© EIOPA 2011
 D. Valuation of assets and liabilities, Own funds and Solvency
Capital Requirement

Valuation of assets and liabilities excluding technical provisions

In Solvency II, assets and liabilities are being valued on a market consistent basis. It
is the aim of Solvency II to make the valuation standards for supervisory purposes, to
the extent possible, compatible with the international accounting developments so as
to limit the administrative burden on undertakings.

QIS5 shows that there continues to be broad support for this economic valuation
approach. Due to the alignment, within the limits of the Solvency II valuation
principles, with international accounting standards, participants from countries where
these standards are in use experienced little difficulty in applying the Solvency II
valuation requirements.

But still, QIS5 has also outlined inconsistent valuations by participants, be it due to
differences from IFRS or the inherent difficulty of applying mark-to-market valuation
for all items. In the former case, participants from countries applying valuation on
amortised cost bases reported more problems and some doubts about the reliability of
the reported QIS5 balance sheet. In the latter case, participants who used mark-to-
model valuation methods did not give much information on the actual techniques
used. In general, small and medium undertakings faced difficulties where the current
accounting basis differs significantly from IFRS.

There is wide variety in the way deferred taxes were recognised and valued. Deferred
taxes seem to be the most difficult area in the valuation of assets and other liabilities,
especially when it comes to recognizing that differed tax assets should be realisable
within a reasonable time frame.

Other areas which have shown inconsistent treatments and different interpretations
are the valuation of intangibles, participations, contingent liabilities, financial liabilities
and employee benefits.

Technical Provisions

There continues to be an overall support, as it was the case with QIS4, for the basic
design of the valuation technical provisions, consisting of the calculation of a best
estimate and a risk margin.

At the same time, a number of areas have been identified that might need further
development:

The Risk Margin calculation, as provided by the full approach, seems overly
complicated, leading to a very large use of the simplifications provided. Further
guidance on simplifications will be needed for ensuring consistency in the
calculation throughout Europe. EIOPA stands ready to undertake such work. As
compared to QIS4, no major concerns have been raised with regards to the
cost of capital factor (6%).

The definition of the contract boundaries seems to be unclear. This leads to

significant differences and a potential unlevel playing field. Further clarification

Page 8 of 153
© EIOPA 2011
 will have to be provided, taking into account to where relevant and appropriate
the work undertaken by the IASB.

QIS5 has provided the first opportunity to test the applicability of an illiquidity
premium to the discount rate used for the calculation of technical provisions.
Three different buckets to which different types of products had to be allocated
have been tested. Supervisors have noted an inconsistent application of the
buckets; either more and consistent guidance or a simple binary approach of
0/100% should be considered. The application of the premium leads to a
reduction of 1% of the technical provisions on average.

With regards to the segmentation by lines of business, in particular the second

level of segmentation for the life business, the added value of this second level
seems limited, particularly when compared to the complexity it adds to the
system.

Own Funds

Participants reported a total amount of available own funds of €921bn. Close to 92%
of this amount (€846bn) has been classified as being the highest quality Tier 1, which
is unrestricted in its use to meet the capital requirements. This high classification of
own funds has been a recurring feature of the previous QIS exercises (QIS3, QIS4).

QIS5 tested the application of Solvency II criteria for basic own funds under the
scenario that no transitional provisions for the recognition of hybrid capital and
subordinated debt would apply. Notwithstanding the identification by participants of
transitional measures as a significant issue, they seem to have been optimistic in
allocating existing hybrid capital and subordinated debt instruments under the
aforementioned scenario. Therefore, the basis for comparing the situations with and
without transitional provisions was undermined by incorrect and incomplete
submissions.

Nevertheless, the amount of subordinated liabilities currently reported at the by QIS5

participants (€48bn at solo level; QIS4 around €42bn; €82bn at group level) gives a
measure of the potential impact of transitional provisions.

QIS5 allowed for the assessment of the reconciliation reserve, which ensures that the
value of all individual basic own fund items is equal to the total of excess of assets
over liabilities and subordinated liabilities. The reconciliation reserve is part of Tier 1
own funds. At EEA level the positive value of the reconciliation reserve (€110bn) is
driven by a reduction in technical provisions (€241bn), offset by a decrease in asset
values and an increase in other liabilities and reduced by the adjustment for expected
profits in future premiums (EPIFP) caused by the separate disclosure of the last item.

The own funds element constituted by expected profits arising from future premiums
(EPIFP) is an important component of own funds, in particular for life and health
insurers. In order to provide a quantification of EPIFP a proxy methodology was
developed for QIS5 in liaison with industry bodies. A total amount of €83.7bn was
reported. The weighted average of EPIFP for those participants that reported EPIFP
amounted to 20% of Tier 1, and in some cases the amount of EPIFP in Tier 1
constituted 50% or more of the own funds, largely accounted for by large
undertakings or groups. However, it needs to be pointed out that only a small number
of participants carried out the calculation and there are wide variations in data among
undertakings and countries. The qualitative comments indicate a range of difficulties
Page 9 of 153
© EIOPA 2011
with the methodology on conceptual, interpretation or practical grounds. Hence these
strong caveats mean the data from the sample cannot be safely extrapolated.

The discussion as to whether the risks attached to future cash flows contributing to
the calculation of technical provisions is sufficiently captured in the capital
requirements is not yet complete. A clear link also exists with the definition of contract
boundaries in the valuation of technical provisions: the broader the contract
boundaries, the more future cash flows would need to be taken into account in the
calculation of technical provisions.

With regard to the adjustments to basic own funds as required by the Solvency II
valuation, QIS5 further analysed the significance of ring-fenced funds on the overall
level of own funds. As already reported in QIS4, the issue is of relevance for a number
of undertakings in certain countries. QIS5 showed progress in the understanding and
quantification of these funds. Nevertheless participants and supervisors would still
need clarification about the identification and treatment of ring-fenced funds, which
should lead to greater consistency in the calculation.

The identification of own funds in excess of the coverage of restricted reserves from
Tier 1 led to the relegation of a significant amount of own funds to Tier 2 (€5.7bn). An
important adjustment to Tier 1 was also made in respect of the deduction of
participations in credit and financial institutions (€18.6bn). Finally, the adjustment for
net deferred tax led to the relegation of an important amount assets from Tier 1 to
Tier 3 (€9.6bn, or 56% of basic Tier 3).

With regard to ancillary own funds, the extent to which undertakings will seek to
make use of these items other than supplementary calls by mutuals once Solvency II
is implemented, remains to be seen. QIS5 provides some perspective on the potential
contribution of ancillary own funds to the own funds, which might assist supervisory
authorities in assessing the likely calls on resources for the approval process. Ancillary
tier 2 own funds (representing items already permitted under Solvency I) amounted
to €11.6bn, concentrated primarily in three countries.

Capital Requirements: SCR standard formula and MCR

There is broad support both from industry and supervisors towards the modular
approach design in Solvency II. Also the aggregation approach has been well
received. The system allows, through the use of correlations among and within the
different modules, for the recognition of diversification effects to acknowledge that all
risks cannot materialize simultaneously.

When looking at the correlations tested, and the changes within the correlations made
by CEIOPS as compared to QIS4, very few comments were received. No major trends
could be identified.

In terms of the composition of the SCR, market risk has the highest weight within the
standard formula, particularly for life undertakings (67%). For non-life the main driver
remains the non-life underwriting risk sub-module (>50%).

Page 10 of 153
© EIOPA 2011
Market Risk

Market risk is the largest component of the standard formula, particularly in life, both
before and after diversification. The main components within the market risk are
equity, spread and interest rate risks.

Following the comments from participants, market risk is still subject to a level of
complexity that could be reduced. The spread risk sub-module has also attracted most
criticism within the market risk module, particularly due to its calibration (considered
either too high or too low) and the complexity especially in the area of structured
products. The look-through approach for structured products, but also for some
investment funds or unit-linked products was deemed to complex by undertakings and
supervisors.

The impact of the concentration risk is in line withy the size of entities, with a higher
impact for small and medium entities.

A large amount of participations were reported by participants (€377bn), in the

majority of cases valued by using the adjusted equity method, but adopting other
mark-to-model valuation mostly for large participations. Participants considered two-
thirds of total participations to be of a strategic nature, attracting the application of a
reduced charge of 22%. QIS5 did not specify the criteria for determining whether a
participation is of strategic nature. Participants responded that in most cases the
degree of control, the long-term nature of the participation as well as the involvement
in the development of activities of the undertaking were considered of key importance
to decide on the strategic nature.

The currency risk module was noted to contain counterintuitive incentives to hold
assets in excess of liabilities in the reporting currency rather than in the currencies of
the underlying liabilities.

Counterparty Default Risk

This module has been most commented upon, mainly with regard to the overly
complex approach tested, which was not felt to be justified in terms of materiality.
Additional work to complete the simplifications already provided in QIS5 should be
carried out to make the module more workable.

In parallel to enhancing simplifications, the treatment of unrated counterparties has

been perceived as disproportionate by undertakings, and more consistency with
regard to the risk charges for the different types of exposures should be aimed for.

Life Underwriting Risk

The main risk drivers of this module include lapse and longevity risk.

This module has been well received both by industry and supervisors. The main area
for improvement was identified in the lapse risk sub-module: not (as was the case in
QIS4) regarding the amount but rather regarding the complexity of the calculation on
a policy-by-policy basis.

Page 11 of 153
© EIOPA 2011
Health Underwriting Risk

This is one of the areas where major changes have been made as compared to the
QIS4 exercise, including among others the introduction of a methodology for health
risk equalisation systems.

For undertakings primarily or solely underwriting health insurance, health

underwriting is the main component in terms of capital requirements, with an average
of 63%.

When looking at the sub-module for health business calculated with techniques similar
to life insurance (SLT), a clear difference with regards to the key risk drivers between
life insurance and health SLT insurance can be observed: disability (76%) is the main
risk driver in health SLT, as compared to longevity and lapse in life. This justifies a
different treatment for the two lines of business.

As it is the case in non-life, the standardised health catastrophe scenarios would

benefit from additional work.

Non-Life Underwriting Risk

For non-life business, the key risk drivers are the number of claims and the potential
mis-estimation of reserves, which are captured in the premium and reserve risk sub-
modules. Lapse risk is a residual risk.

The non-life underwriting risk module has been criticised by industry mainly regarding
the complexity of the catastrophe sub-module. The work on the catastrophe scenarios
is already being carried out by EIOPA together with the industry.

Lapse risk has also been at the centre of attention, more specifically whether the
materiality in some cases justifies keeping the sub-module. However removing this
sub-module may well create wrong incentives in terms of selling practices in non-life.
For that reason, the sub-module is appropriate and should be kept.

As previously indicated, strong concerns have been raised with regard to the
catastrophe scenarios, in respect of the calibration, as well as the complexity and
availability of data. None of the methods proposed was free of concerns, and further
work is needed. This additional work should ensure that catastrophe scenarios are
also suitable to all insurance business, including in particular credit and suretyship
insurance, reinsurance and business written outside the EEA.

Operational Risk

Very few comments were made with regard to operational risk. Nevertheless, the
answers from participants have shown that most undertakings would opt for the
standard formula approach rather than to develop internal models for this specific
risk. There may be different drivers for this trend such as the difficulties to develop
such models (cost, complexity, timing), and this result needs to be viewed in light of
the limited data available in QIS5 on internal models.

Undertaking-specific parameters

As Solvency II is a system designed to incentivise sound risk management,

Undertaking-specific parameters (USP) are seen as a relevant part of such a system,
Page 12 of 153
© EIOPA 2011
which allow, in the areas identified in the QIS5 (premium and reserve risk for non-life
and non-similar to life techniques health, and revision risk for life and similar to life
techniques health business), for replacing the standard formula risk parameters with
parameters specific to undertakings. But for this approach to work, this has to be
done in a sound and consistent manner, avoiding extending USP to all modules of the
standard formula.

Due to time constraints issues or to a lack of data, not sufficient information has been
collected on this area to consider the results to be representative.

There seems to be a clear consensus that USP should not be used to elude the
requirements of partial internal models.

There is also a clear view from supervisors that USP for inflation should not be
allowed, as this would make comparability more complicated.

Risk Mitigation

The Solvency II system is designed to allow for and incentivise risk mitigation
techniques as part of a sound risk management policy. At the same time, it is not
always easy to take this on board in the standard formula without adding too much
unwanted complexity. The calculation of the adjustment for non-proportional
reinsurance summarizes both issues, namely the need to allow for risk mitigation, and
the complexity of the tested adjustment.

Loss absorbing capacity of technical provisions and deferred taxes

The loss absorbing capacity of technical provisions and deferred taxes captures the
extent to which technical provisions would be reduced and deferred taxes would be
affected (decrease of tax liabilities or increase of tax assets) in the event of a shock.

The impact of this loss absorbency is extremely important, potentially decreasing the
BSCR of (re)insurers with more than one third. It should be noted that only around
60% of undertakings who took part in the QIS5 exercise calculated the loss
absorbency adjustments for technical provisions or deferred taxes; this could mean
that the SCR reported in QIS5 may be overstated for the undertakings which did not
perform the calculation.

Another open issue refers to the potential limitations of the loss absorbing capacity of
deferred taxes, and the fact that for groups there may be different tax regimes in
different countries and restrictions on the availability of deferred taxes. EIOPA does
not comment on the specific aspects of the tax regimes as these fall outside the scope
of its mandate. Nevertheless, the aforementioned impact demands the utmost clarity
with regard to methodology and calculation, and EIOPA stands ready, to work in that
direction.

Minimum Capital Requirement

Following QIS4 testing, the MCR is designed on the basis of a combined approach that
incorporates a corridor with a cap (45%) and a floor (25%) referring to the SCR, in
order to ensure the functioning of the supervisory ladder of intervention. The corridor
is complemented with an absolute floor MCR (AMCR) for life, non-life and health
business.

Page 13 of 153
© EIOPA 2011
The system has worked well, without relevant concerns on its functioning, with the
exception of the way the AMCR is articulated for composites, which as required by the
Directive consists of the sum of the AMCR for life and AMCR for non-life, and the
resulting level of it (considered too high).

E. Internal Models

Due to the fact that most internal models have not been finalised yet and because of
the small sample provided, no firm conclusions can be drawn on the comparison of
the size of the capital requirements calculated by internal models and the standard
formula. Furthermore, some undertakings that participated in this exercise are using
internal model techniques which in EIOPA‟s opinion would not yet be in accordance
with the Directive.

Various participants indicated that they would be applying to their supervisory

authority to use an internal model to calculate the Solvency II SCR. Almost all
undertakings that are part of a group (96% of the respondents to the qualitative
questionnaire on internal models) are aiming to use the group internal model to
assess their local SCRs as they consider it better matches their risk profile, subject to
some deviations to adapt to local specificities. In many cases they indicated that they
have already entered into the pre-application phase. However, at the same time,
many of these undertakings have not submitted any qualitative nor quantitative data
regarding their internal model.

In general, it seems that the scope of application of both partial and full internal
models is still subject to some misinterpretations. For example, some participants
reported that their internal model consists of changing only some parameters
compared to the standard formula. Some others asserted that the scope of their
internal model was full, although operational risk was not modelled. In the first case,
the difference between an internal model and the use of undertaking-specific
parameters needs to be upheld. In the second case, it is important to note that a full
internal model should cover all material risks, or else it will be considered as a partial
internal model subject to all relevant requirements.

The modules that the most participants indicate they plan replacing in a partial
internal model are non-life underwriting risk (natural catastrophe risk and premium
and reserve risk), market risk and life underwriting risk.

With regard to internal models‟ tests and standards, QIS5 has shown that the
development stage of participants‟ internal models (group and solo) varies
significantly.

Undertakings were strongly encouraged to provide both standard formula and internal
model data to enable comparisons between these two sets of, calculations. This
included also the alignment of internal model results with the Solvency II standards
(99.5% VaR over one year). Overall, 234 undertakings (about 10% of all participating
undertakings) provided SCR results calculated by using an internal model in QIS5 (29
groups).

Keeping in mind the caveats mentioned above, on average the results still show lower
capital requirements for undertakings intending to use internal models. However, in
some cases results also show requirements that are higher than the standard formula.

Page 14 of 153
© EIOPA 2011
For groups the impact of the use of an internal model seems to further reduce the
capital requirements but similarly, no exact conclusions can be drawn due to the very
small sample provided

F. Groups

Due to the increased participation of groups in QIS5 compared to QIS4, the results
from QIS5 allow for drawing further conclusions to be drawn on the solvency of
groups under Solvency II. For this purpose, QIS5 tested the differences between the
three calculation methods for group solvency foreseen under Solvency II: the
accounting consolidation-based method, the deduction and aggregation method and
the combination of both methods. On average, as can be expected, when groups
applied the deduction and aggregation method, rather than the accounting
consolidation-based method, there was a significantly lower surplus. This is due to the
non-recognition of diversification effects under the former method.

For groups using the accounting consolidation-based method based on the standard
formula a reduction of around €86bn in group surplus compared to Solvency I can be
observed, resulting in a weighted average of QIS5 surplus to Solvency I surplus of
57%. However, the surplus is only reduced by €3bn compared to Solvency I if a
combination of local rules (assuming the use of deduction and aggregation method
with local rules for third countries is allowed), and group (partial) internal models are
used at their current status of development. This impact is particularly material for
large groups.

For groups that submitted internal model results, there was an increase in surplus of
about €6bn moving from Solvency I to QIS5, from €94bn to €100bn. However, it
should be noted that there is a high variability in the results in this area due to the
very small sample of groups that have submitted internal model results.

Similar to the results from QIS4, the group diversification effect is on average equal to
a 20% reduction in group SCR compared to the sum of solo SCRs, naturally varying
among groups depending on the diversity of activities and localisation of the
businesses in the group.

Diversification is caused by two effects. Firstly, capital charges at solo level on intra-
group transactions no longer apply at group level. Secondly “real” diversification
occurs due to more diversified insurance activities of groups compared to solo
undertakings. The impact of the intra-group transactions was eliminated from the
total group diversification effect to assess the “real” diversification benefit; yet this
impact may have been underestimated due to limited reporting on these transactions,
which in turn leads to a potential overestimation of the “real” diversification benefit.
The “real” (i.e. net from intra-group transactions) diversification benefits are mainly
observable in the market and non-life catastrophe modules. Intra-group transactions
mostly impact the capital charges for market (concentration risk and equity risk) and
counterparty default risk, resulting in a reduction of these charges at consolidated
group level compared to the solo charges.

As a default approach, no diversification between entities in the calculation of the risk

margin was allowed. To asses the impact of this calculation, an approximation was
applied, comparing the default approach with a potential diversification in the risk
margin. Overall, the impact of diversification in the risk margin would be
approximately 4% of the group SCR.
Page 15 of 153
© EIOPA 2011
Groups have been asked to consider the availability constraints on own funds at the
group level which prevent the “solo” item from being available to meet the group SCR.
For those firms which identified such constraints, around 8% of their total group own
funds would not be available for covering the group SCR, mainly due to restrictions
related to the existence of restricted surplus funds or ring-fenced funds. In a few
cases, restrictions were related to the location of entities in non-EEA countries.
Contrary to the requirements of the exercise, many groups have included minority
interests in the overall group own funds, hence considering them fully available.

G. Calculation Methods and Practicability issues

One of the aims of QIS5 was to encourage undertakings and supervisors to prepare
for the introduction of Solvency II. By collecting comments on the practicability of the
exercise, EIOPA was able to identify areas where further guidance would be required,
or where the feasibility and complexity of the proposals should be improved in order
to ensure a proper implementation.

QIS5 has shown areas where efforts for reducing the complexity would be welcomed
by participants: the calculation of the counterparty default risk sub-module, the
calculation of the loss absorbing capacity of technical provisions and deferred taxes,
the adjustment for non-proportional reinsurance, the design of non-life and health
CAT risk sub-modules, the calculation of expected profits in future premiums, the
valuation of embedded options and guarantees in contracts, the look-through
approach for structured credit, collective investment schemes or investment funds,
the calculation of the lapse risk module, the application of the contract boundaries and
the illiquidity premium.

To address these issues, EIOPA has identified areas for further work for simplifying
some approaches or developing guidance - see hereunder.

Additional complexity arises from the difficulty experienced by participants in

interpreting the new requirements at this stage of preparation, also in light of parallel
developments, such as for example the accounting standards and the current
uncertainty regarding the proposals for Solvency II implementing measures and Level
3 guidelines and standards.

At the same time, participants have identified areas where they need to make efforts
in the implementation: raising the number and quality of their human resources,
investing in training or improving their data quality and management. These efforts
are still posing challenges and the vast majority of undertakings reported that they
were not yet fully prepared for Solvency II implementation, but that they expected to
be by end 2012.

Participants to QIS5 also mentioned other – non-quantitative – for, which will require
further attention of the industry in preparing for Solvency II, such as governance, risk
management and reporting requirements.

Page 16 of 153
© EIOPA 2011
 H. Key lessons and areas of further work

Following the earlier QIS4 exercise, QIS5 was designed both taking into account a
number of lessons learned from the recent financial crisis, as well as building upon
what worked well in QIS4 (for example the design of the MCR) and what was
perceived as subject to improvement. Among the most relevant changes tested, the
testing of an illiquidity premium, diversification benefits in the risk margin, the
quantification of expected profits in future premiums (EPIFP) or the calculation of
group capital surplus using different calculation methods can be mentioned.

Among the main lessons learned, EIOPA considers that a prudent framework has to be
based upon sound capital and valuation requirements associated with particular
attention to the quality of own funds (based on the Directive criteria for loss
absorbency and permanent availability of such items). This is particularly relevant
when it comes to one of the new tested areas, EPIFP. Any further consideration will
need to take account of the definitional aspects identified during QIS5 as well as
ensuring an outcome that is both economically sound and consistent with the
principles of the Directive.

QIS5 has been conducted bearing in mind that it will be the last opportunity before
the implementation of Solvency II, in January 2013, to undertake such a fully
comprehensive exercise. This implies that further improvements to the system have
to be done on the basis of ad hoc work and tests, rather than by designing a full new
QIS exercise.

EIOPA has identified a number of areas where further work is needed, in order to
improve the functioning of the system, both in terms of enhancing its practicability
and ensuring the right calibration, in line with the Level 1 Directive.

The Solvency II project has been developed and tested for more than ten years, and
QIS exercises are essential tools to ensure that the system is sound and workable. In
particular, QIS5 aimed at testing the following areas:

Design and valuation of the system

The design of Solvency II has received broad support, and this is confirmed by the
results and qualitative comments of QIS5. This is another core lesson from the crisis,
namely that risk-based supervision provides the most appropriate framework for
regulation and supervision. But both consistency and comparability need to be
ensured; and to do so, more work is required in the field of valuation of technical
provisions, including the feasible and economically sound application of an illiquidity
premium, as well as on a consistent definition of contract boundaries, and the
valuation of deferred taxes. Comparability, consistency and a level playing field are
cornerstones of Solvency II. Anti-cyclical mechanisms within the framework, pillar 1
and pillar 2 dampeners,” should help addressing the volatility inherent to Solvency II
valuation rules.

Calibration and impact

QIS5 examined the calibrations in the system. Feedback from the industry and
supervisors suggests that they are generally accepted as appropriate, although some
concrete areas would benefit from further refinement. EIOPA is already working on
some of these areas (e.g. CAT risk and non-life calibration).
Page 17 of 153
© EIOPA 2011
Regarding the impact on the sector, the overall reduction in terms of surplus
does not endanger, based upon the data tested (end 2009), the sustainability
of the sector.

Feasibility and complexity

The huge participation rates and increase of participation of small and medium entities
has provided EIOPA with the perfect benchmark to identify areas where the system
can benefit from more simplicity (either through a less complicated design of the
Standard Formula or via simplifications). EIOPA remains fully committed to advance
towards a system that can be applied by all insurers, yet with sufficient granularity as
to capture all quantitative risks appropriately. The areas to consider include the
calculation of the counterparty default risk sub-module, the calculation of the loss
absorbing capacity of technical provisions and deferred taxes, the design of the non-
life and health catastrophe risk sub-modules, the look-through approach for
structured credit, collective investment schemes or investment funds, the application
of the contract boundaries and the illiquidity premium.

Preparedness

Again, the participation of (re)insurers provides a relatively positive message with

regards to the approach that the sector is taking towards Solvency II. There is work to
be done, particularly with regards to data and internal models, but also very positive
signals been received and welcomed. At supervisory level, Solvency II is also a
learning process: EIOPA and national supervisory authorities are building together the
very much needed expertise. Yet, a lot of work has to be done in this area.

Transition

The intention of Solvency II is to bring risk-based supervision to the field of insurance,

not to disrupt the functioning of undertakings, nor their viability. Transitional
measures, per essence limited, are needed, particularly to ensure sound competition
based on a level playing field and to allow for a smooth transition from Solvency I to
Solvency II. But such measures should not be prolonged in time unduly. Transitional
measures for equivalence with third countries, hybrid capital and subordinated
liabilities, and discount rates on technical provisions make full sense in terms of the
aforementioned objectives.
But we need not only to provide transitional measures, we also need to ensure the
right amount of time and the appropriate scope. Too many transitional measures on
too many topics as well as too much time will strongly disincentivise the shift towards
Solvency II and risk-based supervision or have adverse effects on competition. Too
little time will not help achieving the aforementioned objectives. The review of each
transitional measure will have to indicate whether a transitional measure is still
justified.

Page 18 of 153
© EIOPA 2011
1. Introduction

1.1. Disclaimer

This report sets out the results from the fifth Quantitative Impact Study (QIS5)
conducted by CEIOPS (EIOPA) on the basis of the European Commission‟s Call for
Advice in the framework of the Solvency II project. This impact study was mainly
designed to test the calibration and potential quantitative impact of the proposals
(including a number of alternative approaches), as well as the preparedness of the
industry and supervisors. As such, QIS5 is a field test and not a proposal for the final
Solvency II framework.

Obviously, there remained scope for different interpretations, not least because
Solvency II is a work in progress, and this impacted negatively on the comparability of
the results. This may also explain some of the dispersion between country data, a
phenomenon also found at country level between participants. Undertakings were also
asked to provide results on a „best efforts‟ basis on a relatively short timeline. As a
result the quality of the data is such that detailed analysis has not always been
possible, and all conclusions drawn from it should be seen in that light.

Whenever in this report a reference is made to a statement from a clear minority of

national supervisors (e.g. a reference to „one supervisor‟), this is done because EIOPA
feels it is important to retain as much information from the individual country reports
as possible. When for any issue only the view of a minority of supervisors is given,
this means that the other supervisors did not give an explicit view on this issue.

In many of the comments received by EIOPA it was not explicitly stated whether the
opinions expressed were those of undertakings or their supervisors. References to the
views of “countries” in this report would usually most reasonably be read as
representing the views expressed by undertakings in that country, not necessarily
supported by the supervisor unless that is explicitly stated.

Where reference is made in this report to conclusions relating to the European

insurance sector, please note that this is only insofar as such conclusions can be
drawn from the QIS5 participants. Since QIS5 had a very high participation rate it
does not seem unreasonable to draw such conclusions.

1.2. Background

CEIOPS launched a first QIS (QIS1) in autumn 2005, the results of which were
received in February 2006. The exercise focused on testing the level of prudence in
technical provisions under several hypotheses. In the summer of 2006 CEIOPS
conducted a more comprehensive second impact study (QIS2), which covered both
technical provisions and the calculation of the solvency capital requirement (SCR) and
minimum capital requirement (MCR). QIS2 focused on the methodology of the
solvency requirements; the testing of the calibration of the parameters was left for
the third study (QIS3). Building on the findings of the previous QIS exercises, QIS3
was launched in April 2007. The results of QIS3 were reported in November 2007 and
laid the basis for QIS4, which covered all areas of the proposed regime, including the
balance sheet impact, own funds, and the design and calibration of the standard
formula. QIS4 also looked for the first time at the impact on groups and the

Page 19 of 153
© EIOPA 2011
comparison between internal model and standard formula results. QIS4 results were
published in November 2008.

1.3. Objectives

The results of QIS5 are intended to be of use in the European Commission‟s

development of level 2 implementing measures. To try to ensure that the results
provided a representative view, the target participation rates were significantly
increased from previous QIS exercises. There was a particular emphasis on increasing
participation among small and medium-sized (re)insurance undertakings.

QIS5 aimed to obtain detailed information on the quantitative impact of the proposals
on insurers‟ and reinsurers‟ solvency balance sheets and also to check that the
proposals were aligned with the principles and calibration targets set out in the
Solvency II Framework Directive. It was also the intention to encourage undertakings
and supervisors to prepare for the introduction of Solvency II and identify areas where
further preparatory work may be required, and to provide a starting point for ongoing
dialogue between supervisors and the industry as we move towards Solvency II
implementation. Finally, it would also allow EIOPA to assess the feasibility and
complexity of the proposals.

We also note that the results will make a useful contribution to ongoing work on the
calibration of the non-life and non-SLT health underwriting risk modules.

There were a number of areas where two possible methods were tested with no
default set:
Discounting with or without transitional provisions;
Internal Models and Standard Formula;
Modular approach and single equivalent scenario for the adjustment for the loss
absorbing capacity of technical provisions and deferred taxes;
Consolidation and „deduction and aggregation‟ methods for groups; and
Local rules and Solvency II rules for groups‟ non-EEA entities.

A final key feature of this QIS exercise was that all group results were submitted to
national supervisors and then to a centralised CEIOPS (EIOPA) database, in order to
allow aggregate analysis to be conducted. This approach was adopted in order to have
sufficient sample size to have a good quality of analysis, and it applied to both the
quantitative and qualitative submissions. Analysis was carried out in close co-
operation with group supervisors, thus benefiting both from a consistent view across
the European market, as well as from member state expertise.

Page 20 of 153
© EIOPA 2011
1.4. Participation
77% of the 4753 European
Graph 1: Number of european (re)insurers
(re)insurers supervised by
EIOPA members and
observers at end 2009 will be 4,753
affected by the Solvency II
directive. A thousand existing 3,680
small undertakings are
expected not to fall into the
2,520
scope of the directive.

68% of the affected

(re)insurers participated on a
voluntary basis in the fifth
quantitative impact study.
Supervised Under SII Scope QIS5 participants
167 groups, including major
groups active on a worldwide
basis as well as groups with business concentrated in a few or even a single EEA
market, provided input allowing policy-makers to better understand the impact of the
Solvency II proposals on a consolidated basis.

This report focuses on the quantitative and qualitative responses of the 2520
(re)insurers and 167 groups which provided usable information. The data received
includes significant overlap in cases where quantitative information was received
twice, once from the undertaking as an autonomous entity – referred to as solo in this
report - and once as a member of a group. Group information was collected on a
worldwide basis, including business conducted outside the EEA and non-insurance
business whether regulated (banking activity) or not. In order to avoid repetition,
group-specific findings are covered in a dedicated section while group findings not
materially different from the solo findings are embedded in the relevant solo sections.
In general, references to the EEA position will indicate the results for solo
undertakings unless otherwise stated.

In its Call for Advice, the European Commission set out a target participation rate of
60% of solo undertakings and 75% of groups. Thanks to close cooperation with
European trade associations and long-time stakeholders and the efforts of national
supervisors, support for the Solvency II project has crystallised in the participation of
an impressive number of (re)insurance undertakings and groups. The participation
target has largely been met and all 30 EEA countries are represented in this study.

The overall increase in participation compared to the previous quantitative impact

study is of more than one thousand undertakings, or an increase of 78%.

Page 21 of 153
© EIOPA 2011
Table 1: Participation in QIS51
Total Number QIS5 Of which
number SII affected participants small medium large
Life 888 799 610 291 236 82
Non-life 2,681 1,879 1,284 834 378 72
Reinsurers 203 182 111 72 26 13
Captive 393 353 175 171 4 0
Composite 588 467 336 142 146 48
All 4,753 3,680 2,520 1,511 791 217
of which Health 1,288 749 382 270 94 18
of which Mutuals 1,509 800 454 337 96 21

Table 3: Participation in QIS5 compared to QIS4

QIS5/QIS4 Of which
small medium large
Life 174% 229% 170% 98%
Non-life 187% 253% 139% 87%
Reinsurers 227% 300% 173% 130%
Captive 177% 174% 400%
Composite 149% 161% 154% 112%
All 178% 227% 152% 99%
of which Mutuals 149% 190% 93% 88%

Table 4: Groups participation by type of group

EEA groups EEA groups EEA subgroup(s) Total
without non- with non- EEA of non-EEA respondents
EEA entities entities groups
Sample size 121 41 5 167

The table below shows the number of group participants by size: large groups were
defined as groups with total assets greater than €90bn, small groups as those with
total assets less than €30bn.

It is important to note the high participation rate among small groups, which explains
most of the improvement in the participation rate between QIS4 and QIS5.

Table 5: Groups participation by size

Total Large Medium Small
Sample size 167 17 23 127

1
As in QIS4, classification of solo undertakings by size was done according to the following table.
Table 2: Limits for size classification
Size Non-life insurers Life insurers
Large > €1bn gross written premiums > €10bn gross technical provisions
Medium €0.1bn - €1bn gross written premiums €1bn - €10bn gross technical provisions
Small < €0.1bn gross written premiums < €1bn gross technical provisions

For reinsurers and composite direct insurers which write both non-life business and life business, the size class was
assigned on a discretionary basis in line with the set classification of non-life insurers and life insurers described
above. For instance:
a composite insurer which conducts medium non-life business and small life business was classified at least
medium;
a composite insurer which conducts medium non-life business and medium life business was classified
medium or large.
Health insurers (defined for QIS5 as undertakings with more than 80% of their technical provisions relating to health
business) were given size classifications according to their legal designation as a life, non-life or composite
undertaking.
Page 22 of 153
© EIOPA 2011
2. Overall financial impact

2.1. Overall surplus

Since the previous QIS, which was run on end 2007 accounts, the insurance sector
financial surplus under the current solvency regime has seen a marked decrease in
2008 (of the order of €200bn) - followed by a partial recovery in 2009. This evolution
is largely explained by the impact the financial crisis had on the value of assets owned
by the sector, and on interest rates used to discount liabilities in some countries. At
the end of 2009 the surplus was approximately €500bn.

Graph 2: Evolution of the current regime surplus (€bn)

400% 700
368.7%
350% 600
300% 309.7%
277.2% 500
250%
400
200%
300
150%
200
100%
50% 100
625.7 425.7 491.1
0% 0
2007 2008 2009

Amount Ratio

The Solvency II framework replaces the existing solvency requirement with a set of
two financial requirements: a threshold triggering immediate and ultimate supervisory
action named the Minimum Capital Requirement (MCR) and a higher, risk-sensitive
capital requirement named the Solvency Capital Requirement (SCR).

Subject to supervisory approval, the standard approach to computing the SCR can be
substituted, wholly or in part, by an undertaking‟s own internal modelling of the own
funds needed to support the risks borne, incentivising sound risk management and
rewarding it.

The current regime‟s requirements are based on applying a common set of rules to
existing accounting figures which are prepared differently in different countries,
resulting in non-harmonised outcomes. In sharp contrast with this, the new regime
applies a principles-based harmonised framework from the ground up:

- The starting point of the solvency assessment is a harmonised prudential

balance sheet valued according to Article 75 of the Solvency II directive 2. This
harmonised balance sheet is not necessarily the same as the one in an
undertaking‟s audited accounts.

- The Solvency Capital Requirement is defined as the potential amount of own

funds that would be consumed by unexpected large events whose probability of
occurrence within a one year time frame is 0.5%. This definition based on a

2
http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2009:335:0001:0155:EN:PDF
Page 23 of 153
© EIOPA 2011
 probability measure allows (and sometimes mandates) the replacement of all or
part of the standard formula with an internal model, when this can be shown to
be better able to fulfil the directive requirements in relation to an undertaking‟s
particular risk profile.

- The Minimum Capital Requirement is defined as the potential amount of own

funds that would be consumed by unexpected events whose probability of
occurrence within a one year time frame is 15%. In order to ensure the smooth
functioning of graduated supervisory intervention (often referred to as “the
ladder of intervention”), the linear result produced by the MCR calculation is
bounded between 25% and 45% of the SCR, subject to an absolute minimum.

- The SCR applies at both solo and group level, whereas the MCR only applies at
solo entity level.

In introducing these two levels of capital requirements on top of a fully harmonised

solvency balance sheet, the new regime makes it possible to simultaneously
incentivise sound risk management by putting the onus on the risk-based SCR and
allow through the lower MCR a graduated supervisory response to any worsening in
an undertaking‟s financial position.

The following graph shows the overall quantitative effect of the switch from the
current requirements3 to the two Solvency II capital thresholds as specified under
QIS5.

Graph 3: Current regime and QIS5 surpluses (€bn) (solo)

676.0

476.3

354.6

Solvency I SCR MCR

The financial position of the European insurance sector remains comfortable assessed
against the standard formula SCR4 calculated according to the QIS5 specifications,
with the eligible amount of own funds to cover the SCR/MCR exceeding the regulatory
requirements by around €360bn. This surplus has decreased by c. €120bn compared

3
Please note that in most instances where reference is made to the surplus under Solvency I, this is taken from the
statistical annex of the FSC annual reports: as such it represents the surplus for all insurers in the relevant countries
rather than only QIS5 participants. Therefore while they are useful for observing the overall trends, the figures are not
directly comparable. Since QIS5 participants covered 95% of EEA technical provisions this should not make a material
difference. For the graph which follows, however, this number has been adjusted relative to the QIS5 data and
participants.
4
Please note that throughout section 2, reference to the SCR relates to the standard formula SCR unless stated
otherwise.
Page 24 of 153
© EIOPA 2011
to the current regime. At the same time, the margin before the MCR, the point of
mandatory supervisory intervention, has increased by €200bn.

As anticipated, the Solvency II regime (as tested under QIS5) has created the desired
range for a ladder of supervisory intervention. At market level, the surplus over MCR
is almost twice the surplus over SCR.

The change in solvency ratios is much greater, but less representative of the overall
impact, as both components of the ratio (capital requirements and eligible solvency
elements) are based on fairly different principles in the two regimes.

Table 6: Capital requirements and surplus

Current regime Solvency II
SCR MCR
Solvency ratio 310% 165% 466%
Surplus 476 355 676
Require 227 547 185
ments
Eligible 703 902 861
own funds

Moving from the current valuation framework to the harmonised Solvency II one
triggers some revaluation effects on the asset and liability sides of the balance sheet
which impact the own funds available to meet the regulatory requirements. Moreover,
some off-balance-sheet items may also be counted as available ancillary own funds.
According to their quality, the available own funds components are eligible to meet
either the MCR, the SCR or both. This explains why eligible own funds have increased
compared to the current regime, but by differing amounts depending on the
regulatory threshold concerned.

Graph 4: Distribution of SCR coverage

More than 400% 13.9%

Between 350% and 400% 5.3%

Between 300% and 350% 7.4%

Between 250% and 300% 9.5%

Between 200% and 250% 12.2%

Between 150% and 200% 17.1%

Between 120% and 150% 11.4%

Between 100% and 120% 8.3%

Between 75% and 100% 6.1%

Less than 75% 8.8%

At individual level, 29% of the participating undertakings had SCR coverage between
120% and 200%, around the market average of 165%, while almost half of all
participating undertakings held more than twice their capital requirements.

Page 25 of 153
© EIOPA 2011
15% of undertakings displayed a solvency ratio of less than 100%, with a significant
number of them only just under the threshold. Conversely some were just above the
100% level.

Since it is a risk-based measure and a significant portion of the risks are linked to the
fast-changing nature of the financial markets, the SCR will be a measure whose
precise value will change much more frequently than the annual rhythm of
observation envisaged. This volatility is acknowledged in a regime which allows for
supervisory judgement to be applied at the first trigger point without immediate
action being compulsory for all cases.

At end 2009, 8.8% of the participating undertakings had a solvency ratio that was
sufficiently far below the 100% level to discard the possibility of a measurement error
or the effects of the inherent short-term volatility of the financial markets being the
cause.

A quarter of these undertakings were group members: in their cases this result
indicates a mismatch in risk and capital allocation within the group, which could be
fairly easily addressed either through capital reallocation or intra-group risk transfer.

In a few cases, the revaluation of the balance sheet using Solvency II principles
resulted in negative own funds.

Graph 5: Distribution of SCR coverage by country

At country level, the percentage
of participants with SCR solvency EEA
ratios above 200% varied -----
between 19% and 80%. In one EE
SK
country, no participant had a LI
solvency ratio below 120%. In FI
LT
four other countries, no NL
undertaking had a solvency ratio CZ
below 75%. FR
DE
PT
In the majority of countries, RO
ES
around 10% of undertakings had AT
a solvency position materially LU
NO
lower than the SCR. This group in DK
particular includes a number of CY
IS
small undertakings, which is IT
unsurprising given the high HU
BE
proportion of them among QIS5 BG
participants. SI
UK
LV
SE
IE
MT
PL
GR

75% to 120% Less than 75% 120% to 200% More than 200%

Page 26 of 153
© EIOPA 2011
The distribution of coverage ratios for the MCR displayed a similar pattern to the
findings for the SCR, albeit with the distribution noticeably shifted upward. While 65%
of undertakings can cover more than twice their MCR, 9.4% display a coverage ratio
under 120%.

Graph 6: Distribution of MCR coverage

More than 400% 25.7%

Between 350% and 400% 7.0%

Between 300% and 350% 8.8%

Between 250% and 300% 10.7%

Between 200% and 250% 15.9%

Between 150% and 200% 16.2%

Between 120% and 150% 6.9%

Between 100% and 120% 4.2%

Between 75% and 100% 2.7%

Less than 75% 2.0%

4.6% of participants across Europe were unable to meet the MCR requirement. The
scale of the shortfall among those undertakings is as follows:

Graph 7: Distribution of MCR shortfall

25%
20%
Participants

15%
10%
5%
0%
Negative Own 50% to 100% 25% to 50% 10% to 25% 5% to 10% Less than 5%
funds
Shortfall as % of required MCR

So around a third (1.7% of all participants) have a shortfall of less than 10%;
however, a quarter (1.3% of all participants) have a shortfall greater than 50% of the
MCR. Overall, 0.6% of all participating undertakings had negative own funds
according to the QIS5 valuation principles.

Page 27 of 153
© EIOPA 2011
2.2. Breakdown of the surplus by country

The breakdown of the overall EEA surplus by country is not homogeneous. In thirteen
countries, the surplus assessed against the SCR is greater than the surplus under the
current regime.

Table 7: Surplus by country

Surplus Current SCR MCR
QIS5 476 355 676
QIS5 adjusted 451 395 676
3.5 6.4 10.7
AT
BE 11.0 11.2 17.9
BG 0.2 0.0 0.4
CY 0.4 0.4 0.6
CZ 2.0 2.3 3.5
DE 95.2 118.2 182.7
DK 15.2 11.4 18.8
EE 0.2 0.3 0.4
ES 19.1 11.9 22.0
FI 4.7 3.6 7.9
FR 105.8 81.5 135.4
GR 0.4 0.7 1.6
HU 0.5 1.1 1.8
IE 13.5 4.8 18.4
IS 0.2 0.1 0.2
IT 25.6 38.4 52.7
LI 0.4 0.2 0.4
LT 0.1 0.1 0.2
LU 4.1 4.3 7.7
LV 0.0 0.1 0.1
MT 0.5 0.4 0.8
NL 25.0 17.3 34.0
NO 7.0 3.2 9.0
PL 4.4 7.4 10.8
PT 2.1 1.2 3.0
RO 0.5 0.3 0.6
SE 71.8 32.4 60.8
SE adjusted5 46.8 32.4 60.8
SI 0.2 0.2 0.8
SK 0.8 1.3 1.7
UK 61.8 -5.5 71.1
6 61.8 34.5 71.1
UK adjusted

5
Please note there are two figures for the SE SCR surplus. The second figure is adjusted to compare the QIS5 model
against the Traffic-Light model currently used in this country.

6
Please note there are two figures for the UK SCR surplus. The second figure is adjusted to remove the effect of
certain risk charges on a small number of current intra-group arrangements which are unlikely to remain in place
under Solvency II. This provides a more accurate presentation of the likely UK surplus position under the new regime.

Page 28 of 153
© EIOPA 2011
For two countries, the surplus figures are presented twice, the adjusted figure aiming
to provide a more accurate presentation of the tested proposal impact. Based on the
raw data, the surplus decrease is around 25%, whereas on adjusted figures, the
surplus reduction is around 10-15%.

The above data should be interpreted carefully, especially when undertaking

comparisons between countries. The aggregated raw surplus per country is dependent
on the rate of QIS5 participation and the structure of participants, which varied
between countries. In individual markets, the overall surplus may be heavily
influenced by a few large undertakings, and thus may not be indicative for the
average insurance undertaking.

Only two countries showed a decrease of the overall surplus between Solvency I and
the MCR.

2.3. The main drivers of the surplus changes

Three main drivers explain the changes in the surplus from the current regime to the
Solvency II framework:
- the shift from the current balance sheet to the harmonised Solvency II balance
sheet;
- the shift from the current requirements to the harmonised Solvency II capital
requirements; and
- the differences in the own funds elements allowed to cover the requirements.

The following graph shows the respective influence of these items, splitting the
valuation impacts into positive and negative effects. As this revaluation changes the
amount of own funds compared to the current situation, it also creates deferred tax
assets or liabilities.

Graph 8: Drivers of the surplus changes - EEA

150%

125%

100%

75%

50%

25%

0%
Surplus SI Assets- Assets+ Other valuation TP- TP+ Tax Own funds Capital reqs QIS5
100.0% -37.6% 30.9% -3.5% -14.9% 66.0% -15.2% 9.1% -58.8% 76.0%
-25%

As can be seen from this graph, the relatively moderate overall impact - a decrease in
the range of 20% of the overall surplus - is the result of individual movements in the
main components of much greater magnitude.

Page 29 of 153
© EIOPA 2011
Among these, it can be noted that at end 2009 the upward and downward
revaluations of assets to conform to the harmonised valuation principles almost offset
one another. On the liability side, the removal of the prudence in existing technical
provisions had a far greater impact (66%) than the revaluation upward of some best
estimates (-15%). The net effect of the revaluation of the balance sheet did not solely
result in an increase in the eligible own funds available to cover the solvency
requirements; a portion of it was instead classified as deferred tax liabilities, which
increased substantially.

The increase in capital requirements going from the Required Solvency Margin (RSM)
to the SCR amounted to 59% of the Solvency I surplus, or 43% when adjusted as
described previously. This amounted to a doubling of capital requirements at EEA
level, and was of the same order as the change in valuation of technical provisions.

At individual country or undertaking level, the overall effect of the tested changes
varied materially, depending on both their risk profile and the impact of applying the
common valuation principles to their balance sheet. As an illustration, the following
table shows the relative size of the SCR and the existing RSM by country7.

Graph 9: SCR compared to RSM

0 2 4 6 8

AT: [242]%
BE: [171]%
BG: [330]%
CY: [168]%
CZ: [258]%
DE: [241]%
DK: [156]%
EE: [172]%
ES: [167]%
FI: [376]%
FR: [167]%
GR: [229]%
HU: [272]%
IE: [358]%
IS: [380]%
IT: [162]%
LI: [381]%
LT: [202]%
LU: [303]%
LV: [178]%
MT: [266]%
NL: [206]%
NO: [269]%
PL: [291]%
PT: [160]%
RO: [208]%
SE: [677]%
SE(*): [134]%
SI: [319]%
SK: [229]%
UK: [191]%
QIS5: [211]%

7
For Sweden, the evolution is presented against both the Required Solvency Margin, and the Traffic-Light Model (*).
Page 30 of 153
© EIOPA 2011
2.4. Impact of diversification

To calculate the Solvency II capital requirement, which is defined at the overall SCR
level, the standard formula applies a modular bottom-up approach in which each of
the underlying risk drivers is modelled using the same calibration as that set by the
directive for the overall result. For QIS5, the sum of the individual risks modelled
totalled more than €1300bn.

To acknowledge the fact that the individual risks are not all expected to materialise at
the same time (e.g. a shock on financial markets and a loss on underwriting risks
would not necessarily crystallise at the same time), the standard formula recognises
the benefits of risk diversification through the use of linear correlation techniques. For
QIS5, these diversification benefits amounted to a €466bn reduction in the total risk
charge at solo level.

The last stage in the derivation of the SCR recognises that if risks were to materialise,
part of their cost might be transferred onto policyholders (e.g. through a reduction in
the bonuses attributed to policies with profit participation), and part of the remaining
cost might result in a reduction in the future taxes expected to be paid to tax
authorities. For QIS5, the expected sharing of the cost of risk crystallisation with
policyholders and tax authorities resulted in a €314bn reduction in the own funds
needed.

Graph 10: Impact of diversification and loss-absorbing capacity

Risks
(€1328bn)

Diversification
(-€466bn / -35.1%)

Sharing
(-€314bn / -23.7%)

SCR
(€547bn / 41.2%)

0% 50% 100% 150% 200% 250%

Overall, the final SCR of €547bn is a little above 41% of the sum of individual risks
modelled. Using this overall risk reduction as a basis for calculating the reduction in
individual risks gives a rough idea of the average real risk charges. Using this simple
approach would for example show that while the initial risk loading for listed equity in
QIS5 was 30% of the equity exposure, the final risk capital required was on average
equivalent to a 12.4% capital charge.

This simple approach overlooks the fact that the diversification benefits are not evenly
distributed between risks, but are dispersed between the modules and sub-modules of
the standard formula through a set of correlation matrices that aim to closely match

Page 31 of 153
© EIOPA 2011
observed correlations between the individual risks and sub-risks (e.g. the correlation
between interest rates and equities is not expected to be the same as the correlation
between life expectancy and fire and other damage to property).

In QIS5 a mathematical technique was tested to allocate back the diversification

benefits to underlying risks taking into account the different levels of correlation
assumed (the Single Equivalent Scenario approach); this makes it possible to more
precisely derive the relative weights of the different risks in the final result.

The following graph compares the simple and Single-Equivalent-Scenario-based

approaches to show the SCR‟s sensitivity to the main risks modelled, giving in
brackets the values of the model weights.

Graph 11: Weighting of the main risks in the SCR

Market Equity [24.5%]

Market Spread [15.6%]
Market Interest Rate [9.4%]
NL Premium & Reserve [7.9%]
NL Catastrophe [4.8%]
Life Lapse [4.8%]
Counterparty [5.2%]
Market Property [6.0%]
Life Longevity [2.8%]
Market Currency [3.7%]
Market Illiquidity Premium [1.3%]
Operational risk [5.1%]
Life Expenses [2.2%]
Market Concentration [1.3%]
Health NSLT [1.6%]
Health SLT [1.0%]
Life Mortality [0.5%]
Life Catastrophe [0.7%]
Life Disability [0.3%]
Ring-fenced funds [0.9%]
Health Catastrophe [0.2%]
Intangible assets [0.2%]
NL Lapse [0.0%]
Life Revision [0.0%]

20% 15% 10% 5% 0% 5% 10% 15% 20% 25% 30%

Simple weights Model weights

Under both approaches, the main risk drivers can be seen to be the principle market
sub-risks (equity, spread and interest rates) followed by the principle non-life
underwriting sub-risks (premium and reserve risk and catastrophe risk).

At the sector level some risks appear marginal (intangible assets, non-life lapse or life
revision risks). However they can be more significant for some individual
undertakings.

2.5. SCR coverage

The below graph considers the surplus of eligible own funds over the SCR for all
undertakings in Europe. The horizontal red line divides undertakings into those which
have eligible own funds exceeding their SCR (above), and those which do not have
enough eligible own funds to meet their SCR (below). A surplus of less than -100%
Page 32 of 153
© EIOPA 2011
shows undertakings which have negative own funds. We can see that approximately
85% of undertakings meet their SCR, and a tiny proportion of undertakings fail to
cover their QIS5 liabilities with assets. Around half of the undertakings have enough
eligible own funds to cover their SCR at least twice over.

Graph 12: SCR surplus as % of SCR

900%

700%

500%

300%

100%

-100%

-300%
10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Participants

2.6. MCR coverage

The surplus over final MCR (taking into account the corridor and AMCR, see chapter 7
for further details) was as follows:

Graph 13: MCR surplus as % of MCR

900%

700%

500%

300%

100%

-100%

-300%
8% 18% 28% 39% 49% 59% 69% 79% 90% 100%

Participants

As stated previously, 4.6% of participants across Europe do not meet the MCR
requirements.

Page 33 of 153
© EIOPA 2011
2.7. Group surplus

As summarised in the table below, an €86bn decrease in group surplus eligible own
funds compared to Solvency I can be observed if the accounting consolidation-based
method with the standard formula is used by all the groups in the sample. If group
internal models were approved at their current stage of development and either
equivalence were granted or transitional measures were put in place allowing the use
of local rules for third countries under deduction and aggregation, the overall surplus
would only be reduced by €3bn. Both internal models and the treatment of third
countries had impacts which were individually material and of a similar magnitude.
The table below also shows the split of their impact by size of group.

Table 8: Ratio of surplus under QIS5 to surplus under Solvency I when using internal
models and local rules for third countries
(€bn) Surplus Solvency I Surplus QIS5 Sample size
Results if internal models were approved and/or local rules under D&A were used for
third countries
Large 109.4 129.5 17
Medium 26.7 18.3 21
Small 64.3 49.5 109
All 200.4 197.4 147
Accounting consolidation-based method with standard formula
Large 109.4 54.6 17
Medium 26.7 15.5 21
Small 64.2 43.6 108
All 200.3 113.7 146

Further analysis of the group results can be found in chapter 9.

Page 34 of 153
© EIOPA 2011
3. Valuation of assets and liabilities other than technical
provisions
Outlined below is the composition of the balance sheet, for both solo undertakings and
groups, and under the valuation principles of QIS5 and the current accounting
regime8.

As touched on in section 2.4, for solo undertakings changes in the valuation of assets
between the two regimes have only a limited impact, whereas there is a more
significant drop in liabilities, largely driven by a decrease in technical provisions. Basic
own funds increase (both in absolute terms and as a proportion of the balance sheet)
and are joined by ancillary own funds, which are not included under the current
regime. Groups see a greater drop in the value of assets, and a smaller, but still
significant, fall in the value of liabilities; however they too have a material increase in
own funds.

It can also be observed that the principle asset categories are unit-linked assets,
corporate bonds, sovereign debt and equities. It should also be noted that some
investment funds were reallocated to other asset categories according to the look-
through approach. Overall groups hold proportionally more unit-linked assets and
corporate bonds than solo undertakings, and fewer equities and reinsurance assets.

The more significant changes in asset structure between the two regimes include the
increased proportion made up by unit-linked assets and sovereign bonds under QIS5,
the drop in investment funds, and a decrease in other assets. Note that goodwill
forms part of the current asset structure, but does not appear under QIS5, something
which has a greater impact for groups. Groups also saw a much greater increase in
the proportion made up by corporate bonds than solo undertakings.

The liabilities side of the balance sheet is unsurprisingly dominated by technical

provisions, in particular for life and unit-linked business. We can observe the
introduction of the risk margin under QIS5, discussed in greater detail in section 4.3,
and can also see an increase in the value of deferred tax liabilities (more significant
than the corresponding increase in deferred tax assets) which results from differences
between the QIS5 balance sheet and the one used under the tax regime.

See section 4.1 for analysis of the changes in the valuation of technical provisions,
and section 8 for discussion of the structure of own funds.

8
It should be noted that the valuation bases of current balance sheets may vary significantly between countries.
Page 35 of 153
© EIOPA 2011
 Graph 14: The current balance sheet (solo)
Total assets + 7,456.6 (structure in % of total assets)
19.1% Unit linked
22.5% Corp bonds
19.0% Sovereign
10.3% Equity
4.1% Mortgage
2.1% Property
3.6% Cash
6.6% Reinsurance
6.0% Investment funds
0.2% Deferred tax assets
0.1% Goodwill
6.4% Other
Total liabilities - -6,713.9
-8.9% Non-Life TP
-4.2% Health TP
-46.5% Life TP
-20.6% Unit-linked TP
0.0% Risk margin
-3.2% Short term liabilities
-0.2% Deferred tax liabilities
-3.9% Others
Basic own funds 742.7 10.0%
2.5% Shares and equivalent
2.0% Share premium account
3.7% Retained earnings
2.1% Other reserves
0.7% Subordinated liabilities
1.0% Others

Page 36 of 153
© EIOPA 2011
 Graph 15: The QIS5 balance sheet (solo)
Total assets + 7,432.4 (structure in % of total assets)
20.0% Unit linked
22.7% Corp bonds
20.4% Sovereign
10.8% Equity
4.0% Mortgage
2.8% Property
3.8% Cash
5.9% Reinsurance
3.7% Investment funds
0.3% Deferred tax assets
5.7% Other
Total liabilities - -6,491.2
-6.7% Non-Life TP
-4.0% Health TP
-45.9% Life TP
-20.7% Unit-linked TP
-1.9% Risk margin
-3.0% Short term liabilities
-1.3% Deferred tax liabilities
-3.8% Others
Basic own funds 941.1 12.7%
1.9% Shares and equivalent
1.7% Share premium account
3.2% Retained earnings
-0.4% Asset adjustments
2.1% Liabilities adjustment
1.1% EPIFP
1.5% Other reserves
0.7% Subordinated liabilities
0.8% Others
Ancillary own funds + 11.9 0.2%

Page 37 of 153
© EIOPA 2011
 Graph 14: The current balance sheet (groups)
Total assets + 6,543.1 (structure in % of total assets)
21.7% Unit linked
24.7% Corp bonds
19.8% Sovereign
6.0% Equity
5.8% Mortgage
2.7% Property
2.6% Cash
2.4% Reinsurance
3.3% Investment funds
0.4% Deferred tax assets
1.0% Goodwill
9.7% Other
Total liabilities - -6,166.3
-7.7% Non-Life TP
-2.9% Health TP
-48.6% Life TP
-21.2% Unit-linked TP
0.0% Risk margin
-2.8% Short term liabilities
-0.7% Deferred tax liabilities
-8.0% Others
Basic own funds 376.7 5.8%
1.6% Shares and equivalent
2.5% Share premium account
2.3% Retained earnings
1.4% Other reserves
1.1% Subordinated liabilities
0.8% Others

Page 38 of 153
© EIOPA 2011
 Graph 15: The QIS5 balance sheet (groups)
Total assets + 6,454.9 (structure in % of total assets)
22.1% Unit linked
26.4% Corp bonds
20.6% Sovereign
6.5% Equity
5.7% Mortgage
3.3% Property
3.1% Cash
2.3% Reinsurance
3.4% Investment funds
0.4% Deferred tax assets
6.2% Other
Total liabilities - -5,935.7
-5.9% Non-Life TP
-3.2% Health TP
-48.8% Life TP
-22.1% Unit-linked TP
-1.5% Risk margin
-2.3% Short term liabilities
-1.2% Deferred tax liabilities
-6.9% Others
Basic own funds 519.1 8.0%
1.1% Shares and equivalent
1.8% Share premium account
1.4% Retained earnings
-1.0% Asset adjustments
0.6% Liabilities adjustment
1.0% EPIFP
1.1% Other reserves
1.3% Subordinated liabilities
0.8% Others
Ancillary own funds + 3.1 0.0%

3.1. General
In general the QIS5 economic valuation requirements for assets and other liabilities
were supported and did not cause many problems. Because of the similarity with EU-
endorsed international accounting standards (IFRS) many undertakings have
experience with most of the valuation requirements. This is especially the case for
undertakings that use IFRS or undertakings that use local GAAP in countries where
the local accounting principles are similar to IFRS valuation principles. Countries
where local accounting principles differ significantly from IFRS and where assets are
valued on a cost basis reported more problems and some doubts about the reliability
of the reported QIS5 balance sheet.

Undertakings reported several cases where a mark to market valuation was not
possible, because markets were nonexistent or illiquid. When mark to market
valuation was not possible, a mark to model approach was adopted, or local GAAP
figures or valuation on a cost basis were used. Undertakings did not give much
information on the mark to model techniques used. In cases where local GAAP or
cost-based figures were used, undertakings often mentioned the materiality principle.
Several respondents asked for more guidance on materiality.

Page 39 of 153
© EIOPA 2011
Notwithstanding the general support for the QIS5 valuation principles, undertakings
and supervisors mentioned some balance sheet items where difficulties were
experienced. In most cases it concerned those items where QIS5 requirements
differed from IFRS requirements. The QIS5 restriction on the use of cost-based
approaches also caused difficulties in cases where this is permitted in the financial
statements – for example in the valuation of property.

The valuation of deferred taxes was found to be a very difficult issue. Undertakings
had different ways of dealing with the valuation and with the assessment of whether
the realisation of deferred tax assets would be probable within a reasonable time
frame.

Other items that were often mentioned as being difficult were: intangibles,
participations (where no market value was available), contingent liabilities, financial
liabilities and employee benefits.

Particularly because of the differences between statutory accounting and Solvency II

rules, one supervisor advocated an external audit for the Solvency II balance sheet.

3.2. Impact
Investments form the largest component on the asset side of the balance sheet and
technical provisions the largest component on the liability side.

On an aggregate level, misstatement in the valuation of most of the assets cited as

areas of difficulty above (intangibles, contingent liabilities, financial liabilities and
employee benefits) does not significantly affect the data quality of the total balance
sheet, although there may be an impact if compared with own funds or the SCR.
However this may not be the case for participations and deferred taxes (in part
because of the latter‟s relationship with the SCR as part of the adjustment for loss
absorbing capacity).

3.3. Materiality
Most undertakings considered the accumulated effect of the materiality principle not
to be significant. Many undertakings used materiality concepts to a rather limited
extent.

Some undertakings used the same materiality decisions as in the IFRS balance sheet.
A few undertakings mentioned explicit benchmarks (for example, a percentage of the
balance sheet, own funds or the SCR).

In the case of immaterial items where market values were not available or mark to
model was difficult to apply, assets and other liabilities were valued in line with
current accounting principles or on a historical cost basis.

Particularly where items with a short duration were considered, undertakings argued
that valuation on a historical cost basis did not differ significantly from fair value
valuation, e.g. for loans and financial liabilities.

Page 40 of 153
© EIOPA 2011
3.4. Mark to model
Several undertakings reported using mark to model for investment assets where
reliable market prices were not available. Investment property, property, plant and
equipment, unlisted bonds and equity, structured credit, preference shares, private
equity, investment funds, mortgages, reinsurance recoverables, private loans, certain
derivatives and term deposits were all mentioned in this respect. On the liability side,
financial liabilities were mentioned.

Some undertakings valued participations on a mark to model basis. Three methods for
valuing participations were envisaged under QIS5: market value from quoted prices,
the adjusted equity method and as a last resort other mark to model approaches. For
subsidiaries the first two approaches were the only ones to be used.

The table below shows that for more than half the participations the adjusted equity
method has been applied. However the larger participations have been valued
applying other mark to model approaches. It is not clear why such a significant
number of participations were valued by mark to model. If market prices were not
applicable – as would be the case with subsidiaries held by the participating
undertakings – it is not clear why the adjusted equity approach was not adopted
unless timing issues prevented the gathering of the necessary data.

Table 9: Methods used for the valuation of participations

Method Share of total
Share of total number value of
of participations participations
Market value from quoted active 21% 18%
markets
Adjusted equity method 54% 32%
Mark to model 26% 50%

Undertakings did not give much information on the mark to model methods used
including on their impact or model errors. Some reported using in-house models or
models provided by external parties. In some cases cost-based methods were
mentioned. This is not in line with the QIS5 requirements – unless it can be justified
under the materiality principle or it can be demonstrated that it is a good proxy for
the economic value.

The degree of judgement involved in a mark to model valuation was in some cases
underlined by the fact that some undertakings provided alternative calculations with
changed assumptions.

3.5. Intangible assets

Most intangibles were valued at nil in the QIS5 balance sheet. Some undertakings
valued software as an intangible asset, often justifying this with reference to the
valuation in IAS 38 or local accounting standards. A couple of undertakings recognised
intangibles in respect of renewal rights and customer relationships, consistent with
their audited financial statements.

From the responses it was not always clear which valuation basis was used. In some
cases undertakings referred to using a cost basis, which is not in line with the QIS5
specifications, which require the use of an economic value. For example, just
recognising the development costs as an asset (which was reported by some
undertakings) is not allowed according to QIS5.
Page 41 of 153
© EIOPA 2011
3.6. Deferred Taxes
There was substantial variation in the way deferred taxes were recognised and valued.
Deferred taxes seem to have been the area of greatest difficulty in the valuation of
assets and liabilities other than technical provisions.

The following categories were apparent:

Undertakings that did not calculate deferred tax assets or liabilities at all. This
seems contrary to expectations, because of the valuation differences between
QIS5 and tax regimes.
Undertakings that did not recognise deferred tax assets because of perceived
uncertainty about their realisation. Those undertakings have only reported
deferred tax liabilities.
Undertakings that reported deferred tax assets but did not comment on the
question regarding whether their realisation within a reasonable time frame is
probable.
Undertakings that reported deferred tax assets and commented that the
assessment of whether they could be realised was not possible or made no sense.
Undertakings that reported deferred tax assets and commented that the
assessment of whether they could be realised did not lead to any adjustments.
Undertakings that explicitly assessed whether their deferred tax assets could be
realised and made adjustments accordingly, resulting in a decrease of deferred tax
assets or in not recognising any deferred tax assets at all. Often similar criteria to
IFRS or local GAAP were used for the assessment.

There is evidence that IAS12 was not followed correctly in many cases. Because of
this it is difficult to assess the impact of deferred taxes in the Solvency II regime.

One country advocates the discounting of deferred taxes, on the basis that this would
be better in line with the Solvency II valuation principles.

3.7. Contingent liabilities

For a lot of undertakings, the amount of contingent liabilities was reported to be
immaterial. Several countries reported that undertakings had difficulties with using
the QIS5 valuation methodology at this stage (recognition and valuation need more
analysis). Some undertakings did report contingent liabilities, including among others:
commitments, guarantees, pledges, insurance and non-insurance-related legal cases
and rental contracts.

3.8. Financial liabilities (other than technical provisions)

For a lot of countries financial liabilities did not comprise a significant proportion of
total liabilities. However, for some undertakings that had financial liabilities, there
appeared to be problems with their valuation.

Some undertakings assessed financial liabilities using IFRS principles on a fair value
basis. Many submissions stated that no adjustment was made for own credit risk;
there were no comments on any resulting impact.

Others used cost of purchase for initial recognition and the amortised cost method for
subsequent measurement, with the justification that these liabilities were short-term,
Page 42 of 153
© EIOPA 2011
that they were not prepared to value them differently or that the required valuation
method was not clear.

One country reported that the requirements for the yield curve to be used in the
valuation of financial liabilities (e.g. subordinated debt) were unclear, particularly in
terms of whether or not to include an illiquidity premium.

One country reported that subordinated liabilities were considered without adjustment
for own credit risk, as they were considered part of own funds.

3.9. Pension liabilities

A lot of countries reported that undertakings do not have pension obligations.
Undertakings that did have such obligations used IAS19 or local GAAP for their
valuation. The elimination of the corridor was only mentioned in some cases. Some
insurers explicitly reported not having considered the elimination of the corridor. No
usage of internal economic modelling for the valuation of pension liabilities was
reported explicitly. Several countries mentioned pension liabilities as an area that had
to be discussed further.

3.10. Investment funds

Some undertakings used a look-through approach when reporting assets in
investment funds, others did not. This may have an impact on any interpretation of
the differences between the Solvency I and QIS5 balance sheets and on deferred tax
assets.

Page 43 of 153
© EIOPA 2011
4. Technical provisions
Under Solvency II, the valuation of technical provisions follows the transfer value
principle, under which the value of technical provisions shall correspond to the current
amount the insurer would have to pay if was to transfer its insurance obligations
immediately to another insurer. To achieve a valuation consistent with this principle,
the technical provisions are calculated as a best estimate plus a risk margin. The best
estimate corresponds to the probability-weighted average of future cash-flows, taking
account of the time value of money. The risk margin represents the cost of providing
an amount of eligible own funds equal to the Solvency Capital Requirement necessary
to support the insurance and reinsurance obligations over the lifetime thereof.

However, where future cash flows associated with insurance obligations can be
replicated reliably using financial instruments for which a reliable market value is
observable, the value of technical provisions associated with those future cash flows
shall be determined “as a whole” based on the market value of those financial
instruments. In this case, separate calculations of the best estimate and the risk
margin shall not be required.

“Net technical provisions” refers to technical provisions net of reinsurance

recoverables.

4.1. Comparison with current regime

It is important to emphasise that the quantitative results must be analysed carefully,
as sometimes significant changes in the value of some items on the balance sheet are
not the result of a real change in the value of that item, but instead result from its
reclassification in the QIS5 balance sheet.

Overall gross technical provisions for all lines of business decreased by 1.4% from
Solvency I to QIS5. The main differences between technical provisions under the QIS5
and Solvency I methodologies can be explained by the following:
the use of a new discounting model including the use of an illiquidity premium;
the absence of any surrender floor;
the recognition of future premiums and charges; and
the use of realistic assumptions in the best estimate calculation (i.e. no implicit
prudence margin, although this is partly offset by the inclusion of an explicit
risk margin in addition to the best estimate).
In the valuation of QIS5 liabilities, management actions and policyholders‟ behaviour,
such as lapses, renewals and surrenders, were taken into account.

For life insurance business net technical provisions in QIS5 increased in comparison
with Solvency I. This was mainly caused by the decrease in reinsurance recoverables,
as gross technical provisions in fact showed a slight decrease of 1.0%.

The different interpretations of the contract boundaries definition have led to

inconsistency between undertakings and may also have led to incorrect calculation of
technical provisions.

The graph below shows a comparison of life net provisions for all QIS5 participants
under QIS5 and Solvency I. We note that total net provisions are greater under QIS5
than under Solvency I and that this is an increase of around 3% (for solo

Page 44 of 153
© EIOPA 2011
undertakings). Net provisions for with profit business increased by 8% under the new
regime.

Graph 16: Ratio of QIS5 net provisions to Solvency I net provisions for life
140% obligations

120%

100%

80%
Solos
60%
Groups
40%

20%

0%
with profit linked policies without profit reinsurance total life

For most non-life lines of business net provisions have decreased from Solvency I to
QIS5; gross provisions for non-life decreased by 24.9%. Please note that equalisation
reserves can no longer be included in the technical provisions. The decrease between
Solvency I and QIS5 for non-life business is mainly due to the discounting of future
cash flows, and the exclusion of the implicit safety margin included in technical
provisions through prudent and cautious assumptions, partially offset by the inclusion
of an explicit risk margin. The observed changes could also be partially due to
different segmentations between the two regimes.

Graph 17: Ratio of QIS5 net provisions to Solvency I net provisions for non-life obligations
120%

100%

80%
Solos
60%
Groups
40%

20%

0%
e
es
y
or
y

ss
p
rt

s
s

s
rty

in
lit
lit

in

in
nc
i
po

sh

l lo
ns
ot

re
pe

re

re
bi
bi

ta
ty
rm

ns

e
lia

cia
lia

sis
o

lty
th

rty
xp
re
tra

pr
he

an

l
al
cle

as
le

ua
su

ea

pe
to

r
d
ot

in
ga
ne

as
hi

lh

ro
nd
an

f
e
ve

le

lc
ge

lp
na
us
ag

ta
n

na
or

na
tio
o
io

di

ne

o
iat
ot

da

tio
e

or

rti
cr

la
m

av

op

r
r

po
el

po
he
e,

isc

pr

o
ot

pr
in

n-

pr
m
ar

n-
no
an

n
m

no
no
e
fir

4.2. Discount rate and illiquidity premium

4.2.1. General comment

Based on the amount of the illiquidity premium risk sub-module in the SCR, which
corresponds to a reduction of 65% of the illiquidity premium included in the valuation
of technical provisions, the effect of the introduction of the illiquidity premium in the
valuation of technical provisions in QIS5 can be estimated as being almost 1% of the
value of technical provisions (which represents around 15% of SCR).

Page 45 of 153
© EIOPA 2011
Several countries noted that there were practical difficulties with the illiquidity
premium, for example in calibrating economic scenario generators to varying discount
rates, and felt that further detailed guidance is needed.

Negative forward rates

It was noted that applying the illiquidity premium to spot (rather than forward) rates
led to technical difficulties (negative forward rates). It was noted that negative
forward rates are a technical anomaly which can cause significant calculation
problems, and that this issue needs to be addressed before Solvency
II is implemented.

4.2.2. Illiquidity premium buckets

The graph below shows that the illiquidity premium buckets of 75% and 50% were
most used by undertakings9.

Graph 18: Split of life Best Estimates by illiquidity premium buckets

100%

80%

60%
100% Bucket
75% Bucket
40% 50% Bucket

20%

0%
EEA GP

50% Bucket

The most common products where 50% of the illiquidity premium was used were non-
life in general, unit- and index-linked business, life without profit participation, SLT
(Similar to Life Techniques) health, non-SLT health and reinsurance (both life and
non-life). Some undertakings also used 50% of the illiquidity premium for life
insurance with profit participation, pure savings products and longevity swaps.

75% Bucket

The most common products where 75% of the illiquidity premium was used were life
insurance with profit participation in general, pure savings products, unit- and index-
linked insurance with guarantees, and various types of annuities. Some undertakings
also used the 75% bucket for SLT health, non-life, non-SLT health, life reinsurance,
annuities from non-life, and life insurance without profit participation.

9
The graph does not include non-life provisions, as these were all to be allocated to the 50% bucket.
Page 46 of 153
© EIOPA 2011
100% Bucket

The most common products where 100% of the illiquidity premium was used were
different types of annuities (including annuities from non-life). 100% of the illiquidity
premium was also used for retirement business in run-off, unit-linked insurance, non-
life insurance, and non-SLT health insurance.

Many supervisors reported inconsistent application of the illiquidity premium buckets

across insurance undertakings. It was noted that detailed guidance is needed on what
products attract the illiquidity premium and to what extent. It is, for instance, unclear
how group annuity policies should be treated and whether they should be included
within the 50% or 100% bucket. Also, some undertakings used criteria such as
whether or not annuities are in payment and whether or not there is an option to
lapse within the product to separate the 75% bucket and the 100% bucket. Some
supervisors also highlighted the practical difficulties that undertakings experienced
when applying the illiquidity premium to various buckets. Some countries suggested a
reform to the approach that was tested in QIS5. While one country was explicitly
against its inclusion at all, a number of others questioned the applicability of the
illiquidity premium to all liabilities. Further consideration or guidance was particularly
requested on whether the illiquidity premium is appropriate for unit-linked business
and how hybrid products would be unbundled. Opinions were mixed on the number of
buckets required, with some countries requesting a two-bucket structure of 0 and
100%, and the majority making no comment. One supervisor would prefer to restrict
the application of the illiquidity premium to an approach based on transitional
measures for specific types of insurance business, and to other cases only in stressed
conditions.

4.2.3. Practicability issues

It was noted by one country that the nature of the QIS exercise as a point in time test
meant that any analysis around countercyclicality was limited at best, and also that
further thought is necessary as to the exact calculation of the amount of illiquidity
premium available in the market.

Clarity and consistency are required to adequately determine how the illiquidity
premium should be attached to various types of business, particularly insurance with
profit participation and business which produces negative technical provisions.

4.2.4. Transitional measures

Three countries provided data on the potential impact of transitional measures on
technical provisions. According to the data available, the transitional measures can
have an impact on all types of business (unit-linked, with and without-profit business)
with a magnitude varying from 1% to 7% of the value of technical provisions without
transitional measures, depending on the product and the bucket of the illiquidity
premium. The amount of technical provisions after the potential effect of transitionals
is shown in the graph below as a percentage of pre-transitional technical provisions,
with the total post-transitional technical provisions also given in absolute terms.

Page 47 of 153
© EIOPA 2011
 Graph 19: Impact of the use of transitional
100%
measures on technical provisions
99%
98%
97%
96%
95%
94%
93%
92%
91%
90%
With profit in Unit linked in Other in With profit in Unit linked in Other in Accepted
bucket 50% bucket 50% bucket 50% bucket 75% bucket 100% bucket 100% reinsurance
(€3.3bn) (€143.8bn) (€34.5bn) (€149.2bn) (€4.4bn) (€128.1bn) in bucket
100%
(€28.8bn)

The vast majority of respondents did not see transitionals as material in their market.
Transitional measures were strongly supported by two countries which see them as a
vital measure for their industries, in particular for long-term liabilities. Another
country noted that whilst transitionals would not currently have an effect, they might
do at the Solvency II Directive implementation date. The most common product to
which transitionals were applied was annuities (immediate and deferred), including
bulk annuities.

Many countries did not identify any products which would be eligible for transitional
discount rates, often because such an approach would not be applicable under current
legislation and hence was disregarded. Some supervisors in countries where
transitionals were applied noted that for some undertakings there was uncertainty as
to what products the discount rate transitional is intended to apply to. In particular it
was not clear whether the transitional should only apply to those products in the
100% bucket.

4.3. Risk margin

4.3.1. Practicability
Very few undertakings across Europe reported having used the full calculation
approach for the valuation of the risk margin. Almost all supervisors noted that a full
calculation was often too complex and time-consuming for undertakings.

Undertakings have therefore largely used the proposed simplifications. Because the
calculation was so burdensome, supervisory authorities often supported the use of
simplifications, some of them also pointing out that the relative immateriality of the
risk margin means that it does not justify such difficult calculations.

Page 48 of 153
© EIOPA 2011
The graphs below show the choice of simplification made by undertakings across
Europe.10

Graph 20:

Risk margin, method Non-life obligations

50%
40%
30%
20%
10%
0%

6.Other
4.Duration
calculation

approximation

approximation

5.% BE
1.Full

2.Risks

3.SCR

Graph 21:

Risk margin, method Life obligations

50%
40%
30%
20%
10%
0%
6.Other
4.Duration
calculation

approximation

approximation

5.% BE
1.Full

2.Risks

3.SCR

Some supervisory authorities expressed a concern that the different methods could
give divergent results, possibly leading to opportunities for regulatory arbitrage; some
authorities therefore wondered whether the number of simplifications should be
reduced or whether it would be useful to explicitly give guidance on the choice of
method.

One supervisor questioned the reliability of the simplifications in general. In particular

it was pointed out that some simplifications may give abnormal results for specific
products: this was especially the case for some business where the main assumption
underlying the calculation (that future SCRs for the reference undertaking and best

10
The approaches referred to here were outlined in the QIS5 Technical Specifications as follows:
1. Make a full calculation of all future SCRs without using simplifications.
2. Approximate the individual risks or sub-risks within some or all modules and sub-modules to be used for the
calculation of future SCRs.
3. Approximate the whole SCR for each future year, e.g. by using a proportional approach.
4. Estimate all future SCRs “at once”, e.g. by using an approximation based on the duration approach.
5. Approximate the risk margin by calculating it as a percentage of the best estimate.
Page 49 of 153
© EIOPA 2011
estimates are proportional) did not necessarily hold true (especially in life and health
where future premiums are taken into account).

There is currently also an issue with the use of simplifications where the best estimate
is negative, because all the simplifications from levels 3 to 5 are based on the
assumption that the risk margin is proportional to the best estimate. The supervisors
affected therefore felt it was important to develop simplifications that would still be
robust in the case of negative best estimates.

In the opinion of one country, undertakings should not take into account catastrophe
risk in calculating future SCRs, since catastrophic claims are reported relatively
quickly.

As shown in the graph, for EEA solo undertakings the risk margin is higher as a
proportion of net best estimate for business without profit participation compared to
other lines of business.

For some non-life lines of business the ratio is a bit higher than 10%.

Graph 22: Ratio of risk margin to net technical provisions for life
9% obligations
8%
7%
6%
5%
4% Solo
3%
Group
2%
1%
0%
Total Life with profit linked policies without profit without profit annuities
stemming from
non-life

Graph 23: Ratio of risk margin to net technical provisions for non-life
obligations
14%
12%
10%
8%
6%
4%
2%
Solos
0%
Groups
al

e
s
y
or
es

n
n

ss
p

ns
rt

ns
se

...
lit

in
ilit

nc
...
t io

io

hi
t

po

l lo
ot
ns

ei
ei
To

ia
re
bi
op

en
at

ys

ta
b
ec

yr
r
m

s
pe

av
lia

cia
lia
ns

t
an

sis

y
lth
xp
pr

re
ot

alt
t
r
ex

e,
pe

er
he

an
l
c le

le

as
su

ea
ra
pr

to
t

su

in
op
l

om

d
ot

fin
ga
ca

ne
hi

lh
d
e
e

ar
an

ca
pr
ag

an
m

ve
i

le
ge

na

lm
ed

'c

us

al
co

am

l
it
rs

or

na
tio
eo
m

na
in

ed
i
ke

tio
t
ot

rd

tio
lan

or
via

tio
cr
or

or
op
he

or
el
,a

or
w

op
op
isc
ot

pr

op
e

pr
in

n-

pr
m
d

pr
ar

n-
an

no

n-
m

no
no
e

no
fir

Page 50 of 153
© EIOPA 2011
 Graph 24: Non-hedgeable provisions to gross best estimate for life
120%
100%
80%
60% Solos
40% Groups
20%
0%
with profit linked policies without profit reinsurance annuities
stemming from
non-life

Graph 25:
Graph 25: Non-hedgeable provisions to gross best estimate for non-life

101%
100%
99%
98%
97%
96%
95%
94%
93%
y

ns
ip

s
es
rt

ss
es

s
n

y
n

ty

..
o

in

in
ilit
lit

nc
io

tio

po

sh

tio
l lo
er
s

ns
ot

ei
re

re
bi
en

ct

ta
ab

yr
sa

op

y
ns
rm

pe

ia
cia
lia
te

et

lth

ty
sis
xp

l li
n

av
tra

alt
pr

ex
o

er
pe

he

an
e
le

as
su

a
a
pr

icl

su

e,
he
to

er

op
l
d
m

in
ot
ca

ga
d

in
e

eh

an

ca
n
co

sf

pr
an
m

al
i

le
ge

ar
ed

ag
v

on

al
n

ou
co

l
'

m
it
rs

na
or

tio

am
m

on
ed
in

rti
ne
ke

al
ot

tio
via

rti
rd

po
cr

on
la
or

or

po
a

el
he

ro
w

rti
op
e,

isc

ro
p
ot

po
in

n-

pr

p
m
ar

ro
d

n-
no
an

n
m

p
no
no

n-
e

no
fir

The two graphs above show the percentage of total provisions which could not be
reliably replicated by assets with an observable market value, and which were
calculated as a best estimate plus risk margin. They show that the percentage of
technical provisions calculated as a whole (those which could be reliably replicated by
assets with an observable market value) in QIS5 is almost nil in non-life, while it is
material in life, especially in reinsurance.

4.3.2. Unavoidable market risk

Lots of questions were raised regarding the inclusion of unavoidable market risk for
the calculation of the risk margin. There was no detailed guidance in the technical
specifications on how to interpret and calculate unavoidable market risk.

Almost all non-life undertakings followed the simplifications stating that it is likely that
this unavoidable market risk is nil for them.

Life undertakings often calculated unavoidable market risk when the duration of their
liabilities was longer than the maturity of assets on an active market (often
considered to be 30 years), as hinted by the technical specifications. Many different
approaches were used for the calculation, including the simplification proposed in the
technical specifications, a recalculation of the interest rate sub-module capital charge
to tackle the mismatch.

Page 51 of 153
© EIOPA 2011
Two other examples of unavoidable market risks were quite often quoted by
undertakings:
the illiquidity premium risk for those who use a replicating synthetic portfolio to
value their liabilities; and
the mismatch between this artificial portfolio and a portfolio that could actually
be bought.

Due to the lack of homogeneity among the answers it is however not meaningful to
derive statistical figures for the unavoidable market risk in life business. The majority
agree that there is a need for further clarification on the methods to be used.

4.3.3. Values
At European level, the ratio of risk margin to technical provisions is on average the
following:

Table 10: Breakdown of life technical provisions and risk margin

11
Life TP breakdown
As a
whole/ BE/total RM/total
total TP TP TP RM / BE
Life TP - with-profit 18.64% 79.89% 1.46% 1.83%
Life TP - linked
policies 29.11% 69.32% 1.56% 2.26%
Life TP - without-
profit 12.81% 80.56% 6.64% 8.24%
Life TP – reinsurance 37.21% 58.89% 3.90% 6.63%
Life TP - annuities
stemming from non-
life contracts 11.82% 85.92% 2.26% 2.63%
Total life 21.95% 76.02% 2.03% 2.67%

Table 11: Ratio of risk margin to technical provisions for non-life

Line of business RM / gross TP
Medical expenses 4.72%
Income protection 8.74%
Workers' compensation 8.33%
Motor vehicle liability 5.32%
Other motor 7.28%
Marine, aviation and transport 5.36%
Fire and other damage to property 7.21%
General liability 6.39%
Credit and suretyship 10.56%
Legal expenses 5.70%
Assistance 6.50%
Miscellaneous financial loss 7.63%

11
For some life business (liabilities which could be reliably replicated by assets with an observable market value)
technical provisions were calculated “as a whole” rather than as a best estimate and risk margin. As such the risk
margin/best estimate ratio, relates only to those liabilities not calculated “as a whole”.
Page 52 of 153
© EIOPA 2011
Non-proportional health reins 9.31%
Non-proportional property reins 11.18%
Non-proportional casualty reins 9.02%
Non-proportional marine, aviation,
transport reins 8.17%
Total non-life 6.75%

These aggregated figures may trigger further consideration of the appropriateness of

method 5 (proportion of the best estimate) or at least its calibration, because if this
had been used, it would have given the following results (bearing in mind that method
5 does not explicitly allow for diversification between lines of business):

Table 12: Ratio of method 5 risk margin to technical provisions for non-life
Line of business RM / TP
Medical expenses 7.83%
Income protection 10.71%
Workers' compensation 9.09%
Motor vehicle liability 7.41%
Other motor 3.85%
Marine, aviation and transport 6.98%
Fire and other damage to property 5.21%
General liability 9.09%
Credit and suretyship 8.68%
Legal expenses 5.66%
Assistance 6.98%
Miscellaneous financial loss 13.04%
Non-proportional health reins 14.53%
Non-proportional property reins 6.54%
Non-proportional casualty reins 14.53%
Non-proportional marine, aviation,
transport reins 7.83%
Total non-life 7.83%

4.4. Segmentation

Segmentation of data according to the QIS5 specifications was often not consistent
with the segmentation used for current reporting. Undertakings indicated that current
reporting systems were not always granular enough to allow accurate segmentation.
Many undertakings were unable to make the appropriate changes to their systems in
advance of QIS5. Therefore some undertakings reported using pragmatic approaches
to make the allocation for the purposes of QIS5.

In most countries, undertakings reported that the guidance on segmentation was not
sufficiently clear. This may have led to different interpretations of the required
segmentation. Only a couple of countries reported that undertakings did not mention
any problems with the segmentation.

Page 53 of 153
© EIOPA 2011
A number of life and non-life undertakings indicated that the segmentation of policy
contracts used in QIS5 was difficult or unclear, including the segmentation into
proportional and non-proportional reinsurance treaties.

4.6.1. Life
Almost all countries reported difficulties with the second level of segmentation (death,
survival, disability/morbidity, saving). The main issue is the idea that a contract is
classified according to the main risk driver at inception of the policy and does not
need to be reclassified over the life of the policy. Undertakings were of the view that
this approach materially distorts the picture, as the nature of a policy changes over
time. Some life undertakings indicated their support for segmentation according to the
relevant risk at reporting date.

Moreover undertakings in several countries reported that it was unclear when to

unbundle a contract.
A few countries reported problems with the segmentation of some hybrid contracts
(e.g. unit-linked products combined with guarantees).
4.6.2. Non-life

- miscellaneous - non-proportional
- assistance; 0% financial loss; 2% health reins; 0%
- legal expenses; 2%
- medical expenses; 4%
- income protection;
- credit and suretyship; 5%
2%

- workers'
compensation; 3%

- general liability; 24%

- motor vehicle
liability; 23%

- other motor; 4%
- fire and other
damage to property; - marine, aviation and
15% transport; 5%

Graph 26: Split of non-life net technical provisions for solo

undertakings

The above chart shows that in QIS5 the principal lines of business for non-life
technical provisions were motor and general liability insurance, followed by fire. These
three lines of business between them represent almost two thirds of non-life technical
provisions.

Page 54 of 153
© EIOPA 2011
Undertakings gave some examples of areas where the QIS5 segments were not
sufficiently granular. They stated that risks within some segments can vary from small
individual risks to large industrial risks and that these risks are inherently quite
different. Typical examples given are the following segments:
Fire and other property damage
Marine, aviation and transport
Miscellaneous non-life insurance
Some undertakings commented that it would also be helpful to distinguish between
material losses and personal injuries in the third party liability line of business (LoB).

In some cases it was indicated that it was unclear when non-life obligations became
life obligations (annuities).

In some countries undertakings reported that it was difficult to split motor business
into “motor vehicle liability” and “motor other”. In some cases undertakings were able
to split claims, if these risks were modelled separately. Premiums, however, could not
always be split this way since an individual contract (and premium) covered both
aspects.

4.6.3. Health

Workers' Graph 27: Split of health gross technical

compensation provisions for solo undertakings
5%
Income
protection
11%

Medical expenses
14%

SLT health
70%

In QIS5 most health technical provisions related to Similar to Life Techniques (SLT)
business. Of the non-SLT business, the largest component was medical expenses
followed by income protection.

The criteria for segmenting health business between SLT and non-SLT were
considered open to interpretation, leading to uncertainty in the case of some specific
lines of business.

Some supervisors indicated that the layout of the CEIOPS spreadsheet added to the
uncertainty. The allocation of annuities in payment was also considered to be a
problem by some participants.

Page 55 of 153
© EIOPA 2011
Two countries reported that undertakings found it relatively difficult to correctly
perform the appropriate allocation to the defined lines of business given the
complexity of health business.

In some countries, undertakings had difficulties with the segmentation of accident

insurance into non-SLT health as some “accident” contracts do not fit into the current
health/non-life segmentation.

One supervisor remarked that the technical specifications are somewhat ambiguous
regarding the classification of disability and morbidity business into life and health.
However, they stated that overall this did not lead to a significantly different
assessment of the underlying risk, with the exception of the SCR catastrophe risk
module. It was indicated that further clarification in this regard is needed.

Finally, some supervisors indicated that undertakings running workers‟ compensation

business criticised the QIS5 approach for its lack of clarity on segmentation and its
implementation.

4.5. Contract boundaries

Many participants found the definition of contract boundaries unclear and some even
suggested that the technical specifications, their annexes and the answers provided in
the Q&A procedure were not always consistent. This has led to differences in
application.

Therefore, a majority of supervisory authorities stated there was a need for further
refinements in order to reduce the heterogeneity of practices and to ensure a level-
playing field.

Many industry participants declared themselves to be in favour of applying the

principles set out in the Exposure Draft of the expected IFRS 4 standards, as they
think they better represent the economic value of their portfolios. In some countries,
participants also mentioned that it was possible that the framework tested in QIS5
was not fully aligned with the Level 1 Text in all cases.

Some supervisory authorities referred to specific contracts that are of paramount

importance in their countries. In these cases the implications of widening or restricting
the boundaries are huge and may lead to misrepresentation of obligations towards
policy-holders: unit-linked contracts, savings products, non-life riders to life contracts,
supplementary health, and contracts providing group coverage.

Several supervisory authorities explicitly supported the industry‟s position, whereas

others commented that it was more important to ensure the most efficient risk-based
approach was used, and to avoid cherry-picking by the industry between IFRS and
Solvency II on particular issues, particularly since IASB will not necessarily endorse all
the features of the Exposure Draft.

In general, irrespective of their position on convergence with IFRS, supervisory

authorities shared the view that the principles set out in the technical specifications
may have unintended consequences. Undertakings may grant more benefits or modify
their terms and conditions in order to extend the amount of cash flows considered to
be within the boundaries of profitable contracts. One supervisor also noted a link
between contract boundaries and the illiquidity premium, observing that extending the
Page 56 of 153
© EIOPA 2011
contract boundaries would add further uncertainty to the cash-flow projections used in
the best estimate calculation.

Participants and supervisory authorities also pointed out the difficulty of interpreting
what should be considered an “unlimited ability to amend the premiums or the
benefits under the contract”.

Linked with their desire for greater convergence with the IASB‟s work, many
participants mentioned that the determination of the boundaries of a group contract
should be at the time when a reassessment of risk is possible at individual level,
rather than the time when it is possible at group level.

EIOPA has already provided feedback on this issue in response to IASB‟s Exposure
Draft.

Some supervisory authorities also mentioned practical difficulties encountered by their

countries with the recognition of contract boundaries, especially for non-life business,
mainly because IT systems were not yet able to include tacit renewals in the current
valuation process. Some authorities think it may be useful to add clarifications on
what is meant by “becoming a party” in terms of liability recognition.

Aside from the theoretical concepts, some participants also seem to have struggled to
collect robust data from which to derive reliable assumptions around policyholders‟
future behaviours, in terms of lapse or renewals and in terms of possible evolutions of
the underlying risks.

It is important to note that the contract boundaries question may have consequences
for other parts of the regime, including EPIFP (see own funds section for further
details), SCR lapse risk and the calibration of non-life premium and reserve risk.

4.6. Other feedback

4.6.1. Life technical provisions

The industry reported general problems with respect to resources, experience,

methodologies, limitations on available data and runtimes for stochastic models.
Undertakings indicated that in some cases the generation of scenarios was expensive
or that a large amount of work was required to model a comparatively small liability.
Insurers generally intend to solve these problems by increasing resources and
developing their capabilities and methodologies.

Industry also indicated difficulties with the valuation of options and guarantees, the
modelling of policyholder behaviour and the reflection of management actions.
Availability of software and data was part of the problem. However problems with the
valuation of options and guarantees were not always seen as material.

More advanced problems that were reported were around the appropriate level of
illiquidity premium to be used for the calculation of technical provisions, calibrating
economic conditions for economic scenario generators to negative forward rates, and
limitations in current stochastic valuation models.

Page 57 of 153
© EIOPA 2011
Participants reported general challenges around assumptions and methods regarding:

Assumptions:
Policyholder behaviour
Future discretionary benefits
Future management actions
Catastrophe risk
Validation of input and output from economic scenario generators
Methods:
Valuation of options and guarantees
Stochastic modelling

Supervisors generally reported similar areas of focus to those mentioned by insurers.

Valuation of options and guarantees and assumptions on policyholder behaviour were
of general concern. One country characterised the valuation of options and guarantees
as a black box and stated that it is hard to check the reliability of the best estimate.

Management actions

The impact of management actions was generally reported as less than 2% of total
technical provisions. However in several countries as many as 30 % of undertakings
indicated an impact of more than 5%. Some insurers did not take management
actions into consideration but acknowledged that they may have some impact,
whereas others did not see them having any impact at all. Management actions may
also be limited in some cases due to contractual rules related to the insurance
policies.

Methods

The following methods were used to calculate technical provisions: Monte Carlo
simulations, closed form stochastic approaches and deterministic approaches. The
choice of method applied varied across countries.

It should be noted that EIOPA does not endorse any of these methods as a default. In
principle, under the proportionality assessment process the obligation is on the insurer
to select a method for the calculation of technical provisions which is designed in a
way that adequately captures the underlying risks. Depending on the individual risk
profile of the business, this generally allows a variety of approaches to be taken.

Monte Carlo simulations

In Monte Carlo simulations, in most countries the majority of firms used 1000
simulations. Some countries reported more than 2000 and in a few cases even more
than 5000 simulations.

For those insurers that replied to the question on Monte Carlo error statistics the error
was generally less than 2%, but also in some cases larger than 6%.

Deterministic approach

Historical experience was used in assessing policyholder behaviour. Some

undertakings questioned whether policyholders behaved rationally. In some cases
countries reported the use of an approach developed on a national level.

Page 58 of 153
© EIOPA 2011
In one country surrender option probability was seen to increase with policy age. For
some undertakings deterministic results were found to be comparable to stochastic
results calculated by Monte Carlo method.

Future discretionary benefits (FDB)

As highlighted in previous exercises, the feedback shows that further clarification in

this area is still needed. Although further details were provided in the technical
specifications, several undertakings still experienced difficulties in interpreting the
technicalities of FDB. This could also have affected the results. One country
particularly highlighted the calculation of FDB as an area where undertakings had
experienced serious difficulties and interpretation issues and where the supervisor had
major concerns.

4.6.2. Non-life technical provisions

Most countries do not expect radical changes in the methodologies adopted by

undertakings for the valuation of non-life claims provisions, but rather an adaptation
of those methodologies to the specificities of the Solvency II framework. Some
supervisors noted that it is difficult to assess the reliability or adequacy of the results.
This would require more detailed information on the assumptions and parameters
used and on the quality of the underlying data, which was not feasible in the
framework of the QIS exercise.

It appears that particular attention needs to be given to the valuation of premium

provisions and the recognition of catastrophe (CAT) claims. Several countries
indicated that it would be helpful to have further guidance on how to calculate the
best estimate for premium provisions, with attention paid to the treatment of
acquisition expenses and other costs, as well as the derivation of the combined ratio
per line of business (LoB).

Graph 28: Ratio of gross claims provision to gross BE for non-life obligations
120%
100%
80%
60%
40%
20% Solos
0% Groups
y

ip

ce

ns
s
l

rt

ss
es

or
n

s
n

ty
ta

..
in
se

in
ilit
lit
io

tio

po

sh

tio
l lo
er
s

ot

an

ei
To

re

re
bi

en
en

ct

ab

yr
sa

op

ty
ns
rm

ia
ist

ia
lia
te

ty
xp
xp

re
l li
n

av
tra

alt
pr

nc

t
s
o

er
pe

l
he
e
le

le

as
su

ea
ra
pr

a
icl

su

e,
to

p
d
om

fin
ot
ica

ne

ga

lh

ro
d

in
e

eh

an

ca
e

an
m

le
ge

ar
na

lp
ed

'c

ag

us
v

al
on
co

lm
it
rs

na
or

tio
eo
am
m

on
ed
in

i
ke

t
ot

io

na
or
n
ia

rti
rd

cr

la
or

t
av

op

tio
or

o
el
he
w

op
op
pr
e,

isc

or
ot

pr
in

n-

pr

op
m
ar

n-
no

pr
an

n
m

no
no

n-
e

no
fir

Premium provisions

Most undertakings reported using proxy techniques to calculate premium provisions.

The second simplification described in the technical specifications (expected claims
ratio) has been widely used.

Page 59 of 153
© EIOPA 2011
Most undertakings expressed a desire to improve on the methods used in QIS5.
Some participants took the view that the method based on the expected claims ratio
which is described in the technical specifications is already sufficiently risk-sensitive
and did not see the need to develop more sophisticated approaches.
Given that in most cases calculation of the premium provision through projection of
cash flows was not possible due to the lack of appropriate historical data, the most
important next step which undertakings are planning to take is collecting more data
and enhancing the quality and granularity of data available, with special regard to
information about segmentation of products, lapse options, future premiums,
expenses, binary events and CAT claims. In addition, several undertakings noted a
need to improve and clarify the treatment of contract boundaries.

Claims provisions

As reported in previous exercises, run-off triangles were widely used by undertakings

in the determination of the best estimate of claims provisions. Generally the chain-
ladder or Bornhuetter-Ferguson methodology was applied, occasionally with
adjustments for claims inflation.

Other methods mentioned by some undertakings were Mack, Fisher Lange, the
stochastic method, using the tool provided by CEIOPS, the expected claims ratio
method, the method by Hodes, Feldblum & Blumsohn, and the Benktander method.

The most common techniques adopted by undertakings to calculate claims provisions

were:
Chain-ladder techniques based on paid claims, claims incurred or number of
claims;
Bornhuetter-Ferguson techniques based on paid claims or claims incurred;
De Vylder least squares;
Loss ratio methods;
Stochastic, for example bootstrap or Mack methods; and
Frequency/severity analysis.

Often these techniques were used to derive best estimate provisions gross of
reinsurance. In such cases, amounts net of reinsurance were determined using one of
the Gross-to-Net proxies provided in the specifications or similar techniques.

Claims which had been reported but not yet settled, particularly large claims and
claims of an exceptional nature, were dealt with on a case-by-case basis by
undertakings in many countries. Actuarial judgment was applied to determine the
most appropriate method.

Most of the participants do not report any plans for enhancement of the methods that
were used in QIS5, except that the methods must be improved to include the effects
of inflation. Some undertakings reported an intention to introduce stochastic models.

4.7. Reinsurance recoverables

The graph below shows a comparison between the best estimate (BE) for recoverables
and the gross best estimate for provisions in non-life, and indicates that for the
marine, aviation and transport (MAT) line of business the best estimate for
recoverables represents approximately 45% of the total gross best estimate for
provisions. Several other lines of business show ratios greater than 30%.

Page 60 of 153
© EIOPA 2011
 Graph 29: Ratio of BE for recoverables to total gross BE for non-life
50%
45%
40%
35%
30%
25%
20%
15%
10%
5% Solos
0% Groups

an nce
l

s
su ility
r
cle tion
pe ion

an atio the lity

s
l e hip
to t

pr rtio ope ins

ta

...
to

or
ke e p nse

se

tio l he l los

ar r...
tio casu r...
ed era ...
To

av
e
ot n an mo

p
bi

en
om c t

no rop us f sista
p

ab
ve nsa

ty

no opo al p th r
am ans
pe

l m lty
rty
cia
lia
' c ot e

e,
p
l li

re
r
x

in
al
tr

s
in al e

e
r

a
ag
d

in
a
hi

d
o

r
ic

g
n
om

an

le
ge

a
ed

l
na
op ion

na
rd

it
rs

or
c

o
m

n
ne

t
ot

he

or
cr

no ella
or

or
m

or
av
w

op
d
e,

isc

pr
p
pr
in

n-
m

n-

n-
ar

n
no
fir
m

Graph 30: Ratio of claims provision for recoverables to BE for recoverables for
140% non-life
120%
100%
80%
60%
40%
20%
0% Solos
Groups
y

ip

ns
s
l

ss
es

or
n

s
n

ty
ta

or

in
se

in
ilit
lit

nc

.
io

tio

sh

l lo

t..
er
s

ot

ei
To

re

re
sp
bi

en
en

ct

ta
ab

ia
yr
sa

op

ty
m

cia
lia
te

an

th

rty
sis
xp
xp

av
re
l li
n

t
pr
r
o

l
pe

al
he

an
r
e

ua
le

le

as
su

pe
ra
pr

e,
icl

he
to
d
om

in
ot
ca

ne

ga

as

in
ro
d
e

eh

an

sf
an

ar
m

al

lc
i

le
ge

lp
ed

'c

ag
v

on
n

u
co

lm
na
it
rs

na
or

io

eo
am
m

ed
in

tio
ke

na
t
ot

tio
or
n
ia

rd

cr

la
or

or

tio
av

op

or
el
he
w

op

or
op
pr
e,

isc
ot

pr
in

op
n-

pr
m
ar

n-
no

pr
an

n
m

no
no

n-
e

no
fir

Graph 31: Ratio of BE for recoverables to total gross BE for life

35%
30%
25%
20%
Solo
15%
Groups
10%
5%
0%
with profit linked policies without profit reinsurance annuities stemming
from non-life

Participants reported encountering difficulties with calculating the probability of

expected default of counterparties. This was either because it was difficult to
understand what was specified in QIS5 or because undertakings did not agree with
the specifications. In many cases undertakings reported a lack of data, which for
instance ruled out the calculation of run-off triangles.

Page 61 of 153
© EIOPA 2011
Most participants calculated their reinsurance recoverables by subtracting the net best
estimates from the gross best estimates. Few participants conducted cash flow
projections and simulations.

It was commented that the net best estimate per line of business could be hard to
calculate in some cases, when a single reinsurance treaty dealt with several lines of
business.

Most countries indicated that special purpose vehicles are of minor importance for
most undertakings.

Supervisors indicated that determining the unadjusted best estimate reinsurance

recoverables did not seem to present any particular challenges. However there was
more uncertainty around the calculation of the expected counterparty default
adjustment, with extensive reliance on rating agency assessments for probability of
default.

Some supervisors indicated that it was difficult to check whether the adjustment for
expected counterparty default had been properly taken into account. Others indicated
that there was a concern around the lack of rating for certain third-country reinsurers.

Page 62 of 153
© EIOPA 2011
5. SCR – Standard formula

5.1. The overall SCR

The Solvency Capital Requirement (SCR) is the risk-based capital requirement for
undertakings under Solvency II. It is calibrated to a 99.5% Value at Risk confidence
level over one year. In structure the SCR is composed of a number of „modules‟ which
in turn are composed of „sub-modules‟. The capital requirements arising from these
sub-modules and modules are aggregated using a correlation matrix.

Composition of the SCR

The chart below shows the composition of the SCR for Europe as a whole, first for solo
undertakings and then for groups.

Graph 32: BSCR structure (solo)

200%
180%
160% 30% 32%
8% 0% 8%
140%
28% 57%
120%
12%
100% 1%
80%
148%
60%
102% 100%
40%
20%
0%
Adj TP/DT
Life

RFF
Non-Life
Market

SCR
Intangible

BSCR
Counterparty

Health

Diversification

SCR Op

Diversified risk charge

Page 63 of 153
© EIOPA 2011
Graph 32: BSCR structure (groups)

250%

200%
31% 46%
10% 10%
150% 0%
34%
60%
8%
100% 3%
149%
50% 113% 100%

0%

Adj TP/DT
Life

RFF
Non-Life
Market

SCR
Intangible

BSCR
Counterparty

Health

Diversification

SCR Op
Diversified risk charge

This graph masks significant differences between different types of business, table
xSCR1 in the annex shows those distributions.

Diversification benefit is an important component of the SCR, and the below table
divides the diversification benefit between the risk modules.

The diversification at group level is understandably higher as groups generally include

entities conducting a varied range of activities and thereby generating more
diversification. For detailed analysis of the impact of diversification on groups please
see section 9.3.

Page 64 of 153
© EIOPA 2011
Graph 33: Diversified BSCR structure - All undertakings (solo)
120%

100% 0.2%
16.9%
80% 4.3%
15.3%
60% 6.9%
100.0%
40%

56.5%
20%

0%
Market Counterparty Life Health Non-Life Intangible BSCR

Diversified risk charge

Graph 33: Diversified BSCR structure - All undertakings (groups)

120%

100% 0.3%
15.8%
80% 4.9%
17.3%
60% 3.9%
100.0%
40%
57.8%
20%

0%
Market Counterparty Life Health Non-Life Intangible BSCR

Diversified risk charge

Page 65 of 153
© EIOPA 2011
 Graph 34: SCR Structure - All undertakings (solo)
120%

100% 5.4%
38.7%
80%
0.9%
60%
100.0%
40%
67.4%
20%

0%
BSCR SCR Op Adj TP/DT RFF SCR

Diversified risk charge

Graph 34: SCR Structure - All undertakings (groups)

120%

100% 6.4%
40.1%
80%
2.1%
60%
100.0%
40%
67.0%
20%

0%
BSCR SCR Op Adj TP/DT RFF SCR

Diversified risk charge

The charts below reproduce this analysis for undertakings which write predominantly
life business and undertakings which write predominantly non-life business. As would
be expected, life undertakings have very little underwriting risk arising from anything
other than life but relatively more market risk, while in the case of non-life
undertakings the most significant risk is non-life underwriting and the share of market
risk is smaller.

Page 66 of 153
© EIOPA 2011
Graph 35: Diversified BSCR - Life undertakings (solo)

120%

100% 1.0% 0.0% 0.1%

23.7%
80%
7.7%
60%
100.0%
40%
67.4%
20%

0%
Market Counterparty Life Health Non-Life Intangible BSCR

Diversified risk charge

Graph 36: Diversified BSCR - Non-life undertakings (solo)

120%

100% 0.4%

80%
52.4%
60%
100.0%
7.0%
40% 0.5%
7.0%

20%
32.8%

0%
Market Counterparty Life Health Non-Life Intangible BSCR

Diversified risk charge

Page 67 of 153
© EIOPA 2011
Major SCR Issues

Some parts of the standard formula SCR led to little comment from undertakings
(such as life underwriting and aggregation). Other areas had more comments,
particularly around difficulty of calculation.

Whilst these areas are covered in considerable detail in the remainder of the report,
the following provides a brief summary:
Non-life catastrophe risk attracted comments on methods, calibration, data
availability and the effort required to calculate the risk charge.
Counterparty default risk attracted significant comment on the difficulty of
applying the full calculation and whether the methods are proportional to its
relative lack of importance for many undertakings.
The correct calculation of loss absorbency of deferred taxes caused problems
for some.
The equivalent scenario was less widely-used than the modular approach, and
where it was used there was greater uncertainty around the results. Almost all
countries reported shortcomings with the method on both complexity and more
theoretical grounds.
Lapse risk caused difficulties for both life and non-life undertakings with life
undertakings strongly objecting to the requirement to model the risk at policy
level, and non-life undertakings noting that in many cases they did not have
systems and processes in place to model the risk.
The look-through test proved difficult for some, with guidance requested on the
application of proportionality.

Risk mitigation techniques are covered separately below as they span a number of
areas within the standard formula. There was a general theme of difficulties with risk
mitigation in relation to counterparty default risk, catastrophe risk, and other areas.

5.2. Single equivalent scenario methodology

The QIS5 Technical Specifications defined the single equivalent scenario as the default
method for determining the SCR. However, only 39%12 of participating undertakings
completed the calculation. Feedback from supervisors indicated that small and
medium-sized undertakings in particular omitted the calculation.

This lack of engagement with the method was accompanied by extensive feedback
from industry, as well as from supervisors, on its shortcomings. Almost all countries
reported complaints from their industries on the complexity and impracticability of the
single equivalent scenario. In addition, there seem to have been issues with the
stability of the approach. However, no authority elaborated on the latter any further
than attributing it to general input sensitivity.

Complexity issues first and foremost included a lack of proper understanding of the
method by the industry, but also covered a number of other concerns. The guidance
given in the technical specifications on the single equivalent scenario was limited, as
was the time the industry had at its disposal to complete QIS5.

12
For the purposes of this statistic, undertakings for which the ratio of the adjustment for technical provisions with
the equivalent scenario and the modular approach was between 99% and 101% (i.e. the results were identical or near
identical) have been considered as not having completed the equivalent scenario calculation.
Page 68 of 153
© EIOPA 2011
Undertakings completing the equivalent scenario reported that the complexity it
added to the standard formula did not pay off in terms of deeper insight. Difficulties
applying it to composite undertakings were particularly highlighted, since the model
did not provide a method for consolidating different lines of business.

Since one of the key aims of introducing the methodology to the standard formula was
to streamline the adjustments for future discretionary benefits (FDB) and deferred
taxes, it is notable that a considerable number of reported issues relate to these
adjustments. Countries elaborating on this reported that shortcuts were taken by the
industry in the equivalent scenario calculation of the FDB adjustment (such as
proportional reduction of the modular adjustment), to be able to complete the
calculation within the given time frame.

In summary, it can be said that the single equivalent scenario approach was rejected
by a large majority of participants and supervisory authorities, in most cases on the
basis of the increase in the complexity of the standard formula, which was perceived
as unjustified and overly burdensome. However, a couple of authorities nonetheless
acknowledged the technical appropriateness of the method.

5.3. Loss absorbing capacity of technical provisions and

deferred taxes

The reduction in the SCR coming from loss absorbing capacity stems from the
undertaking‟s ability either to reduce payments of discretionary benefits (loss
absorbing capacity of technical provisions) or to pay less tax than initially expected
(loss absorbing capacity of deferred taxes) after an adverse event. More precisely, the
loss absorbing capacity of technical provisions and deferred taxes captures the extent
to which technical provisions and deferred tax liabilities would be reduced in the event
of a shock.

Loss absorbency is extremely material to the total SCR, as set out in the charts in
section 5.1. The table below shows a number of statistics on the calculation of loss
absorbency. We can observe the loss absorbency calculated under both the equivalent
scenario and the modular approach, and the amount of future discretionary benefits
(FDB) consumed by the loss absorbing capacity of technical provisions under the
equivalent scenario (ES), and the modular approach (MA).

At group level, the adjustment for the loss absorbing capacity of technical provisions
allowed for an overall reduction of 28% of basic SCR and benefited about half of the
participating groups.

The ratio of FDB to group SCR varied considerably between groups, due to the various
different group structures since FDB only exists for life or health SLT contracts.

Proper assessment of the adjustment for the loss absorbing capacity of technical
provisions was not easy at group level as any solo level constraints had to be
assessed and dealt with carefully so as not to result in any undue compensation.
Some participants (both groups and solo undertakings) reported difficulties with
interpreting the FDB definition and with calculating the adjustment.

The adjustment for deferred taxes was on average 19% for groups. The calculation
was again reported to be more complex at group level than at solo level, as groups
often carry out business in more than one country and as a consequence deal with
Page 69 of 153
© EIOPA 2011
different taxation regimes. Moreover, some fiscal regulation provides for the
possibility of fiscal integration at group level. Furthermore at group level deferred
taxes are an item that needs to be taken into account carefully when assessing
availability of group own funds at group level (see group section).

Table 13: Comparison of the equivalent scenario and modular approaches

Weighted Standard Sample
10th 25th 50th 75th 90th average Deviation Size
Equiv Scenario /
Modular Approach 78% 100% 100% 100% 100% 90% 61% 1434
Equiv Scenario /
Modular Approach - TP 85% 94% 100% 100% 128% 98% 62% 285
Equiv Scenario /
Modular Approach - DT 100% 100% 100% 100% 100% 92% 17% 1340
Loss Absorbing
Capacity TP / FDB: ES -92% -74% -44% -20% -14% -43% 35% 290
Loss Absorbing
Capacity TP / FDB: MA -85% -63% -35% -18% -7% -42% 33% 506

The first three lines of the above table show the ratio of the adjustment calculated by
the equivalent scenario, to the one calculated using the modular approach – for the
total combined adjustment in the first line, and then for technical provisions and
deferred taxes separately. This shows that on average the equivalent scenario gave a
slightly smaller adjustment than the modular approach. However we note that for a
large number of firms the equivalent scenario and modular approach results were
exactly equal, suggesting that some undertakings may have given the modular
approach result for both rather than undertaking the equivalent scenario calculation.

The bottom two lines of the table then show the calculated reduction in future
discretionary benefits in the event of an adverse shock. For both the weighted
average was a reduction of around 40%.

Only around 60% of the undertakings who took part in QIS5 calculated a loss
absorbency adjustment, which may mean that the SCR is overstated for undertakings
which did not perform the calculation. This, together with industries‟ and supervisors‟
calls for it, indicates a strong need for additional technical guidance.

One supervisor highlighted that undertakings had experienced serious difficulties and
interpretation issues with the calculation of the loss-absorbing capacity of technical
provisions and cited this as an area of major concern. A majority of countries reported
experiencing difficulties with the methodology for calculating the adjustment for the
loss absorbency of deferred taxes, calling for additional guidance in this area. Many
reported that due to the lack of clarity around the methodology, a variety of
approaches had been adopted by undertakings, or that some had not undertaken the
calculation at all. Some supervisors suggested that this may have had a major impact
on the results.

Some supervisors expressed a concern that qualitative assessment of whether

deferred tax assets can be realised within a reasonable time frame has not been
properly taken into account when assessing their loss absorbing capacity, and also
raised concerns around the subjectivity of this decision. However, another supervisor
strongly argued for full loss absorbency unless there is evidence in a given jurisdiction
that this is not the case. Some supervisors reported on the other hand that their
industry had adopted the QIS4 approach of capping the deferred tax adjustment at
the value of existing deferred tax liabilities, mainly for reasons of practicability. In
addition, in one country some undertakings struggled with the calculation of deferred
Page 70 of 153
© EIOPA 2011
taxes as they knew the local tax regime was likely to change in response to Solvency
II.

5.4. SCR Aggregation and operational risk

The aggregation methodology was generally well received, with no major or

widespread complaints. A minority of undertakings were concerned that the
correlation matrix approach would not adequately capture the effects of non-linearity
and tail dependence. It was noted by supervisors that though there are a number of
limitations to the SCR aggregation approach (see CP74), they still feel it appropriate
for the purposes of the standard formula, and that its calibration is fitting for a 99.5
VaR measure.

A few undertakings commented that the “tiered” aggregation structure was

inappropriate. For example, the method is unable to accurately reflect the interactions
between sub-modules belonging to separate risk modules (the method implicitly
assumes the same correlation between equity and lapse as between equity and
mortality), although again supervisors considered the application as it stands
adequate. A minority of undertakings complained about the two-sided correlation
matrix for market risk (for interest rate risk) since there would be increased
complexity due to additional volatility of results over time. Only a few individual
undertakings which provided internal model input made comments on parameters
used in their correlation matrixes.

Qualitative feedback on operational risk was scarce and mainly focused on the method
being too crude and not giving adequate incentives for good risk management
practices. In this light it is surprising that most undertakings which plan to use partial
internal models indicated an intention to use the standard formula methodology to
assess their operational risk. Operational risk will often simply be added to the other
risks without diversification, as in the standard formula. Groups also intended to use
the standard formula for operational risk due to a lack of data and in the awareness
that it lacks risk-sensitivity.

5.5. Market risk

5.5.1. General market risk

Market risk is the largest component of the standard formula SCR for the European
industry. The equity, spread and interest rate components are the largest elements
within this module, although the relative importance of the sub-modules varies widely
by type of undertaking.

Page 71 of 153
© EIOPA 2011
Graph 37: Market Risk Composition (solo)
160%
140%
6% 8%
120% 10% 36%
100% 30%
80% 12%
60%
42% 100%
40%
20%
28%
0%
Interest Rate Equity Property Spread Currency Concentration Illiquidity DiversificationMarket Risk
Premium

Graph 37: Market Risk Composition (groups)

160%
140% 10%
5%
120% 13% 40%
100%
43%
80%
60% 14%
100%
40% 35%
20%
20%
0%
Interest Rate Equity Property Spread Currency Concentration Illiquidity DiversificationMarket Risk
Premium

Perhaps the greatest area of comment on market risk was spread risk where
comments were made around the calibration of the module, and issues around
complexity for structured products and consistency of charge across different types of
credit-risky assets (government bonds, corporate bonds, covered bonds and
securitisations).

Following this were concerns about the application of the „look-through‟ test for unit-
linked business, and particularly where it would be appropriate to apply
proportionality.

Almost all of the other sub-modules attracted some comments, which are described in
more detail below.

As well as comments on specific sub-modules, there were some more general

comments on market risk:

1. Undertakings and supervisors from a few countries felt that the absence of
equity and interest rate volatility stresses was a significant omission from the
standard formula resulting in perverse risk management incentives, although

Page 72 of 153
© EIOPA 2011
 the majority of supervisors did not raise this point. In most countries, volatility
was one of the major additional risks included in internal models.

2. In addition to this there were comments on the lack of recognition for

geographical diversification within an asset class and the fact that the ratings-
based approach to certain risks penalised undertakings in lower-rated countries.

5.5.2. Look-through approach

A key issue related to market risk was the application of the look-through approach to
unit-linked business. Many undertakings found this extremely time-consuming, and
disproportionate to the (often second order) magnitude of the market risk related to
unit-linked business. A number of supervisors noted that they are supportive of the
principle of substance over form in general, but that guidance is needed in this case.

A significant number of countries saw scope for simplifications in the look-through

approach used in the market risk module, particularly for investments in unit-linked
funds. For more details refer to chapter 10 on practicability and preparedness. Others
requested that the principle of proportionality be spelled out in clearer detail to
understand when it is proportionate to apply the look-through approach.

Where structured products were discussed, there were some comments that whilst
the look-through test is a desirable principle, it has been very hard to implement
according to the technical specification for securitisations and undertakings have
occasionally used ad hoc simplifications. One country explicitly requested
simplifications for structured products.

5.5.3. Interest rate risk

In terms of the interest rate risk sub-module, two countries stated that its calibration
proved to be penal for their industry, especially in view of its interplay with the
current methodology to derive the interest rate term structure, which was considered
inconsistent by one. A couple of countries regarded it as too complex and called for
simplifications, in most cases based on durations (refer to chapter 10 on practicability
and preparedness for further details). One supervisory authority also commented that
basing the charge on the highest net interest rate risk led to inconsistencies with local
regulations, and suggested using the highest overall SCR instead.

5.5.4. Equity risk

From two countries there was feedback that the equity risk sub-module was too penal
for assets backing long-term liabilities, such as retirement insurance or third party
liabilities insurance. Another country regarded the sub-module as over-calibrated
overall and the same country commented that the „other equity‟ charge for Plant and
Equipment was excessive. Some commented on the lack of an equity volatility charge
as a missing key component of the sub-module.

See section 5.12 for details of the equity risk charges applied to participations.

See below the composition of „other equity‟ as reported in the qualitative

questionnaire. The principle components are private equity and hedge fund exposures,
as well as non-EEA exposures not covered by the principle categories. A large
proportion is still classified as “other”.
Page 73 of 153
© EIOPA 2011
 Graph 38: 'Other' Equity by category

16%

28%

Hedge Fund
Private Equity
Commodity
Non EEA
2% 26%
Infrastructure
Innov EU Proj
7%
Other

2%
19%

5.5.5. Property risk

Property risk did not attract widespread comment. In some countries participants fed
back that the property risk module was insufficiently granular, suggesting that
location of property and type of use be taken into account. However in most of those
cases their supervisors stated that they were reluctant to introduce further complexity
to the standard formula by increasing the granularity of this module. In addition, in
three countries there was feedback that the module was overly severe for the local
market. Finally, it was suggested by one supervisor that the calibration was
inappropriate for assets backing long-term liabilities.

5.5.6. Currency risk

Again there was little widespread comment on currency risk. In a couple of countries
undertakings felt that currency risk was overestimated, and two noted a
counterintuitive incentive to hold the reporting currency rather than the currency of
underlying liabilities. Two countries with pegs to the Euro regarded the shock as too
severe given their pegs.

5.5.7. Spread risk

The most commented on sub-module of market risk was spread risk, around which
there were various concerns, falling into three broad areas: calibration, consistency
and complexity.

Page 74 of 153
© EIOPA 2011
On calibration, undertakings in two countries found the spread risk sub-module to
have too high a calibration although one other country considered that the deviation
from CEIOPS‟ advice has led to an overly low calibration. There were also comments
that the sub-module was over-calibrated for structured credit (where in some cases
capital charges were almost 100%) and local government bonds specifically.

On consistency, a few countries expressed the view that the non-inclusion of spread
risk on EEA sovereign debt led to the omission of a risk and skewed incentives for
undertakings. There were a couple of comments on the special treatment for covered
bonds, with one country requesting that the special treatment be extended to AA
bonds. Finally, one country commented that the duration floor of one year was
inappropriate when applied to term deposits of lesser terms.

Apart from that, the complexity of the module is of concern for a couple of countries,
particularly as it relates to structured products. Even the simplifications offered by the
technical specifications are considered to be too complex by some. For further details
on suggested simplifications please refer to section 10.5 (SCR simplifications).

Charts xSCR9-11 in the annex show the distribution by rating type of credit-risky
assets held across the EEA industry, as reported in the Assets tab of the QIS5
spreadsheet. We note that the Assets tab was not particularly fully completed, and
that this data therefore captures only a subset of the market.

One supervisor noted that in their country the industry has significant exposure to
residential mortgages. Therefore both industry and supervisor emphasised the need to
include both exposure to residential mortgages and the risk-mitigating effect of the
National Guarantee Scheme for residential mortgages in the spread risk or
counterparty default risk module. Both would prefer a methodology in line with the
Basel framework.

5.5.8. Concentration risk

Concentration risk was also commented on by few countries, with the impression
being that the sub-module is broadly appropriate.

Another raised concerns around the impact of the sub-module for countries with fewer
banks, and in particular fewer banks with higher ratings. The same country
commented that the treatment of intra-group term deposits and intra-group
investments in financial holding companies for concentration risk was inconsistent,
and that Plant and Equipment should not be subject to this sub-module.

The below graph shows the pre-diversification proportion of market risk made up by
concentration risk for small, medium and large undertakings. As we would expect this
shows that concentration risk can be very material for the smaller undertakings.

Page 75 of 153
© EIOPA 2011
 Graph 39: Concentration Risk / Market Risk by size of undertaking

40%
35%
30% Solo
25%
20% Group
15%
10%
5%
0%
Large Medium Small

5.5.9. Illiquidity premium risk

Some undertakings saw the illiquidity premium risk sub-module as inappropriate or
unnecessary. One country noted that this shock only referred to the impact on the
liability side of the balance sheet, neglecting the assets, and suggested that it be
redesigned to take this into account and penalise undertakings with assets and
liabilities ill-matched in illiquidity terms.

These comments have to be interpreted in light of the fact that the negative
correlation of the illiquidity premium risk sub-module with the spread risk sub-module
can be analytically shown to reduce the overall risk charge in the market risk module,
whenever spread risk charge is bigger than the illiquidity premium risk. As shown in
chapter 2, the marginal effect of the illiquidity premium in the SCR is quite limited
(1.3%).

5.6. Counterparty Default Risk

The feedback from countries on the practicability of the implementation of

counterparty default risk within the standard formula was quite unanimous. The
calculations demanded by the module were widely perceived as being extremely
laborious and complex, especially in view of the fact that the charge demanded for
counterparty risk by the SCR standard formula is quite limited.

The main criticism with regard to complexity was directed at the determination of the
risk-mitigating effect for type 1 exposures. In this respect, two major issues can be
identified:
1. The determination of the risk-mitigating effect on single counterparty level is
perceived as being disproportionately burdensome.
2. Practicability problems with the treatment of reinsurance in other parts of the
standard formula emerge again in the counterparty default risk module when
determining the risk-mitigating effect for a reinsurance counterparty. Problems
were reported in relation to coinsurance pools, derivatives backing life
obligations, reinsurance programmes including more than one counterparty,
and non-proportional reinsurance such as stop-loss treaties.

In addition to that, complexity issues also arose from the cross-dependency of

catastrophe and counterparty default risk, namely in the segmentation of formula
SCR.6.29 per line of business, since the corresponding non-life CAT perils may not be

Page 76 of 153
© EIOPA 2011
limited to one line of business and the technical specifications did not offer a method
for the disaggregation of CAT charges. One country raised concerns that the combined
structure of the CAT risk and counterparty default risk modules essentially assumed
that reinsurers would default following a catastrophe event, which was unrealistic and
unduly burdensome.

Many supervisors supported their industries‟ call for simplification of the module, some
of them explicitly supporting the simplifications already offered by the technical
specifications.

But even the simplifications offered by the QIS5 Technical Specifications were
regarded as too complex by a number of participants. The entire setup of the module
was regarded as disproportionately onerous relative to the nature, scale and
complexity of this risk to undertakings‟ business. A considerable number of
participants suggested simplifications (see section 10.5 on SCR simplifications).

A number of countries reported their industries‟ criticism of the treatment of unrated

counterparties, which was perceived as being disproportionate. In this context,
intragroup-transactions, counterparties with no rating but a positive experience of
past transactions, premium debtors, and hospitals were mentioned. Also, the risk
charge for type 2 exposures was perceived as being disproportionate compared to
type 1 by undertakings in a number of countries. In this context, the 3 month limit for
past-due exposures was also mentioned as being judged as too restrictive for certain
transactions (e.g. receivables for intermediaries). Also, differentiation towards the risk
horizon of derivatives was called for, as well as discrimination between OTC- and ETD-
derivatives. Some countries reported the lack of ratings for counterparties in the
domestic market as being an issue.

In addition to that, a couple of consistency issues with regard to the risk charge for
cash at bank were indicated:
- Undertakings reported the risk charge for cash at bank in the counterparty
module to be significantly higher than the charge for a bond issued by the same
bank in the spread risk module.
- The same holds true for cash at bank in comparison to long-term deposits,
which according to the Q&A were attributed to the spread risk module. In this
context, guidance on the distinction between cash at bank and bank deposits
was also called for.
- In contrast to reinsurance contracts and derivatives, the loss given default
(LGD) for cash at bank bears no recovery rate in the counterparty module.

5.7. Life Underwriting risk

Life underwriting risk is the second most material module for life undertakings behind
market risk. Within this lapse risk and longevity risk are the two most material sub-
modules.

Page 77 of 153
© EIOPA 2011
Graph 40: Life Underwriting Risk Composition - All undertakings (solo)
160%
140%
0% 11%
120% 36%
23%
100%
80% 49%
60%
6% 100%
40%
36%
20%
0%
11%
Mortality Longevity Disability Lapse Expenses Revision CatastropheDiversification Net Life
Underwriting
Risk

Graph 40: Life Underwriting Risk Composition - All undertakings (groups)

180%
160%
14%
140% 0%
25% 55%
120%
100%
56%
80%
60% 6% 100%
40% 32%
20%
22%
0%
Mortality Longevity Disability Lapse Expenses Revision CatastropheDiversification Net Life
Underwriting
Risk

Graph 41: Life Underwriting Risk Composition - Life undertakings (solo)

160%
140%
0% 8%
120% 22% 35%
100%
80% 46%
60% 6% 100%
40%
44%
20%
0% 8%
Mortality Longevity Disability Lapse Expenses Revision CatastropheDiversification Net Life
Underwriting
Risk

Page 78 of 153
© EIOPA 2011
Life underwriting risk has been generally well received, and the impression is of a
module that most of the industry is content with. The only major exception to this is
lapse risk, where there were concerns by many on the requirement to calculate lapse
on a policy-by-policy rather than model point basis.

5.7.1. Lapse risk

The key practical criticism was the need to calculate lapses on a policy-by-policy
basis: a large number of undertakings raised this as an area in need of simplification,
and generally they were supported by their supervisors. Criticisms were that this was
too onerous in terms of calculation time (especially for complex or stochastic models)
and that new systems will have to be developed at significant cost. In some cases ad
hoc simplifications were performed, or model points were used.

There were also criticisms from undertakings and some supervisors of the policy-by-
policy approach on more theoretical grounds, with some suggesting that the
treatment of surrender strain should not be asymmetric and should be by broad
segment to better reflect lack of policyholder rationality. A minority of countries noted
that the asymmetric treatment is appropriate, and some that policy-by-policy
modelling is appropriate for certain types of products, although mentioning that
proportionality should apply.

Some said that taking the maximum of the three shocks was insufficient, and that a
more subtle approach should be applied, and some questioned the dividing line
between wholesale and retail business, usually because it caught the wrong business
(rather than due to the concept of the division itself). A number found segmentation
by surrender strain type very difficult.

There were some other practical considerations raised by one or two countries, with
one reporting that the calculation of lapse effects on options proved challenging.
Another remarked that understanding the direction of the surrender strain of a policy
was tricky and that the module could be simplified by omitting this distinction. Some
requested that lapse penalties be taken into account more explicitly as without this
the current approach may give the wrong „biting‟ lapse stress. There were
inconsistencies noted with applying the lapse to guaranteed annuity options, and
further guidance was requested, with one undertaking requesting that the decrease in
election/take up rates be stressed, and some noting that clarification on the timing of
the surrender should be given.

5.7.2. Concerns in other sub-modules

The principle area of comment other than lapse risk was longevity, and the mortality
risk sub-module also attracted a few items of feedback: further details can be found
below. However neither is considered a major area of discrepancy or difficulty. There
were various comments that particular modules were over-calibrated; however, in
each case this was only from a small minority of respondents.

Undertakings in some countries found policy-by-policy calculation onerous for the

other areas of life underwriting as well as lapse risk. A few countries encountered
difficulties with the unbundling of the different risks, and some others had data
availability problems. There were also a couple of comments that diversification
benefits should be allowed.
Page 79 of 153
© EIOPA 2011
Longevity risk

There was feedback from a number of countries that as the current shock was only a
shock on the level, it failed to adequately take into account trend risk: undertakings
felt a stress on the future improvement rates would be more appropriate. However
opinion among their supervisors was mixed: some agreed that this shock would be
more appropriate, but there were also concerns that this would introduce further
complexity to the standard formula.

Additionally, one country suggested that the positive correlation between lapse and
longevity risks was inappropriate for annuities business.

Mortality risk

One country suggested that the classification of policies between mortality and
longevity should be done based on the stress scenario rather than the base scenario,
as they might show different characteristics in the stress situation. A couple of others
suggested that more compensation should be allowed for mortality and longevity
policies in the same sub-segment.

5.8. Health underwriting risk

The health underwriting risk module had been subject to a complete overhaul since
QIS4, and hence attracted a considerable number of comments. Key areas of concern
were segmentation, the disability/morbidity sub-module, lapse risk, and catastrophe
risk.

Many countries reported that their industries had problems properly segmenting their
health business into SLT (Similar to Life Techniques) and non-SLT (Not Similar to Life
Techniques) lines. One area especially mentioned in this context was workers‟
compensation. In addition to that, two countries reported problems with unbundling
income protection and medical expenses.

Data analysis reveals that for undertakings primarily writing health business health
underwriting risk constitutes a major part of the overall capital requirement, the
proportion of net health capital requirement to BSCR averaging 63% (see graph 4213
below). Hence, even though health underwriting contributes only 4.3% to the overall
EEA SCR (see graph 33 in section 5.1), it is of major importance for the 382 health
undertakings which participated.

13
In graph 42, “mainly health” refers to undertakings classified as “health undertakings” for the purposes of QIS5
(e.g. having more than 80% of technical provisions made up by health lines of business), while the remaining
categories relate to other undertakings which had health liabilities, but for whom it was less than 80% of their
business.
Page 80 of 153
© EIOPA 2011
 Graph 42: Health Risk as a proportion of BSCR

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%
All With Health Risk Mainly Health Life Some Health Non Life Some Health Composite Some
Health

10%ile to 90%ile Inter Quartile Median Mean

The largest portion of the risk charge for health underwriting in the EEA relates to the
non-SLT health underwriting risk module.

Graph 43: Health Underwriting Risk Composition (solo)

120%

100% 11% 8%

80%
59%
60%
100%
40%

20% 39%

0%
SLT NSLT Catastrophe Diversification Gross Health
Underwriting Risk

Page 81 of 153
© EIOPA 2011
Graph 43: Health Underwriting Risk Composition (groups)

120%

100%
9% 10%
80%

60% 55%
100%
40%

20% 36%

0%
SLT NSLT Catastrophe Diversification Gross Health
Underwriting Risk

5.8.1. Health SLT

Feedback on health SLT concentrated on the two sub-modules contributing the most
to the risk charge, namely the disability/morbidity (76%) and lapse risk (19%) sub-
modules.

Graph 44: Health SLT Underwriting Risk Composition (solo)

140%
120% 7%
11% 26%
100% 19%
80%
60%
76% 100%
40%
20%
11%
0% 3%
Mortality Longevity Disability Lapse Expenses Revision Diversification Health SLT
Underwriting
Risk

Page 82 of 153
© EIOPA 2011
Graph 44: Health SLT Underwriting Risk Composition (groups)
140%
9% 4%
120% 31%
100% 27%
80%
60%
79% 100%
40%
20%
3% 10%
0%
Mortality Longevity Disability Lapse Expenses Revision Diversification Health SLT
Underwriting
Risk

On the mortality and longevity sub-modules only one country volunteered comments,
asking for a reduction in the complexity of the approach in countries where selection
by individual life expectancy does not take place and suggesting a portfolio approach.
The revision risk sub-module also received few comments, with one country flagging
the inconsistency of applying different shocks to annuities stemming from health and
non-life.

Disability/morbidity risk is the heavyweight of the health SLT underwriting risk

module, contributing 76% of the risk charge before diversification. A number of
countries reported that their industry regarded the shocks for income protection in the
disability/morbidity sub-module as too severe. One country also reported technical
issues with the calculation, indicating that its health SLT business is not based on
disability/morbidity rates and hence the shocks could not be meaningfully applied,
with the industry using proxy solutions instead. This country also indicated that the
one-sided stresses are counterintuitive for premium adjustment business with profit
participation, since the safety loading included in the premiums may technically lead
to profits in the long run and to a risk charge of nil.

On health SLT Lapse, complexity as well as consistency issues can be identified.

Regarding complexity, it seems to be difficult for the industry to identify the positive
and negative surrender strains required by the module, and the disaggregation of
model points to policy level is also reported as problematic. Regarding consistency,
some countries expressed concerns about the fact that the lapse calibration varies
between the health SLT underwriting and life underwriting sub-modules, commenting
that the relevant contracts may be very similar or that contracts may combine life and
health components. However, one supervisor explicitly supported the distinction, since
SLT health lapse is subject to considerable legal constraints in its market.

5.8.2. Non-SLT health

Comments on non-SLT health focused mainly on the lapse sub-module, and indicated
that the industry in some countries regarded it as immaterial and hence many
undertakings did not calculate it. Countries did not give much feedback on premium
and reserve risk. Two supervisors remarked on difficulties with taking more than one
non-proportional reinsurance treaty into account properly.
Page 83 of 153
© EIOPA 2011
5.8.3. Health equalisation systems
The volatility parameter for health underwriting risk allows for the risk-mitigating
effect of a health risk equalisation system. With the introduction of a HRES-parameter
the volatility parameter can be adjusted, up to an upper level of 50% of the Europe-
wide calibration.

One supervisor indicated that in QIS5 the health insurers‟ risk was better represented
than in previous exercises as the methodology of a health risk equalisation system
(HRES) has now been better accounted for.

The health insurers of this country applied a parameter adjusted by the full 50%. This
meant an increase in the volatility level, as even with the cap this meant a doubling of
the prescribed volatility in comparison with QIS4.

Health undertakings reported that the inclusion of the HRES methodology better
reflected their underlying risks. However they also expressed concerns about the
calibration of the HRES-parameter.

The supervisor also indicated that health insurers‟ risk was better represented with
the inclusion of the HRES in the framework. They advise that use of the HRES-
parameter should be facilitated in situations where the risk equalisation system is still
developing.

The ongoing recalibration of the HRES parameter by the Joint Working Group for the
calibration of non-life and health underwriting risk factors may lead to additional
insights.

5.8.4. Health catastrophe risk sub-module

Graph 45: Health CAT composition (solo)

120.0%

100.0% 7%
19%

80.0%

60.0%
62%
100%
40.0%

20.0%
26%
0.0%
Arena Pandemic Concentration Diversification Health Cat
Diversified risk charge

Page 84 of 153
© EIOPA 2011
 Graph 45: Health CAT composition (groups)

120.0%

14%
100.0% 25%

80.0%

60.0%
71%
100%
40.0%

20.0%
19%
0.0%
Arena Pandemic Concentration Diversification Health Cat
Diversified risk charge

The main comments on health catastrophe risk focused on the appropriateness of the
scenarios to the local market and individual insurers. Hence, this risk module was
regarded as too severe for some undertakings and to be ignorant of certain risk
events for others. Some examples mentioned by countries:
no account is taken of medical expenses, which is an important risk driver in
case of pandemic;
if a country has a singular large stadium it has a significant impact; the average
size of a country‟s stadia rather than the largest stadium should be considered;
the assumption that all three scenarios will take place in the next year is
unrealistic, only the most appropriate should be considered; and
the capital-at-risk is considered more than once in calculating the capital charge
for accidental death, medical expenses, and short-term disability.

Two countries suggested there was double-counting, for example saying that
pandemic scenarios are implicitly included in the disability/morbidity sub-module or
that the arena scenario overlaps with the concentration scenario.

There were also a number of comments that the health catastrophe sub-module was
excessively complex and that data requirements were hard to meet; in particular a
significant number of countries remarked that total insured lives was unavailable for
certain areas.

Two countries questioned the appropriateness of the scenarios in light of the role
played by the public health-care system in a catastrophe situation. The health CAT
risk was criticised by undertakings in a few countries because the concentration
scenario did not allow for different numbers of policyholders for different products,
was arbitrary and could lead to an unlevel playing field. It was also emphasised that
there is an inconsistency between the CAT modules for health and life, as also pointed
out by the CAT Task Force.

Page 85 of 153
© EIOPA 2011
5.9. Non-life Underwriting risk

The non-life underwriting risk module received a lot of criticism regarding complexity
from participants. Very little of this was around premium and reserve risk, however,
with the catastrophe risk sub-module clearly topping the overall list of complaints. In
addition, the lapse risk sub-module was perceived as being immaterial by a large
proportion of participants, and hence the effort involved in calculating the stress was
judged by many to be superfluous.

In addition to that, there was some feedback from undertakings that while the
introduction of future premiums and contract boundaries made sense from a
theoretical point of view, the difficulties encountered in calculating them outweighed
the benefits.

A few countries reported that the correlations were regarded as inaccurate, but did
not identify any specific problem areas. There were also one or two comments that
the basis of the calibrations in historical data could lead to them being inappropriate.

Undertakings in a couple of countries encountered difficulties with the allocation into

lines of business (LoBs), and undertakings in a few countries would welcome
additional guidance on the definitions of written and earned premiums.

Graph 46: Non-Life Underwriting Risk Composition (solo)

140%

120% 1%
20%
100%
50%
80%

60%
100%
40%
70%
20%

0%
Premium & Reserve Catastrophe Lapse Diversification Gross Non-Life
Underwriting Risk

Page 86 of 153
© EIOPA 2011
 Graph 46: Non-Life Underwriting Risk Composition (groups)
140%

120% 0%
20%
100% 41%

80%

60%
100%
40% 79%

20%

0%
Premium & Reserve Catastrophe Lapse Diversification Gross Non-Life
Underwriting Risk

5.9.1. Premium and reserve risk

There were not many comments on the premium and reserve risk sub-module. One
concern raised by undertakings in some countries was around the volume measure
used, as it included commission and hence resulted in a higher risk charge for more
profitable business. There was also feedback from a couple of countries that it was
difficult to omit catastrophes from historical data for the purposes of this sub-module,
but most did not report any difficulties.

There were some comments that geographical diversification was not accurately
reflected, but no clear common view on the precise problem: a couple of countries
suggested that the 25% limit was too low for some insurers, and another that there
should be a single European region. One supervisor noted that the calculation could
be difficult for reinsurers, as data was not always available.

5.9.2. Lapse risk

Most countries reported that this sub-module was found to be immaterial or irrelevant
by non-life undertakings, and in some countries undertakings had omitted the
calculation entirely. The link with the definition of contract boundaries should also be
highlighted; if these were very widely defined, the risk could become more material. A
number of countries commented that the complexity of the calculation had been seen
as unjustified in comparison with the sub-module‟s materiality. Difficulties in
calculation or in sourcing data were not reported by a significant number of countries,
but it should be borne in mind that in some other countries undertakings had not
carried out the calculation at all.

Beyond that, there was some feedback along similar lines to the corresponding life
sub-module: that the policy-by-policy approach was challenging, and that the
policyholder rationality assumed was unreasonable.

Page 87 of 153
© EIOPA 2011
A couple of countries also reported that for reinsurers it was unclear whether the lapse
referred to the reinsurance policies, or the underlying primary insurance policies. In
the latter case it was noted that data might not be available.

5.9.3. Catastrophe risk

Graph 47: Non-life CAT composition (solo)

120%

100% 4%

80% 48%

60% 11%
30% 100%
40%
56%
20% 37%

0%
NatCAT Man-made Natural/man- Total Method Method 2 Method Total CAT risk
made 1 1/method 2
diversification diversification

Graph 48: Natural catastrophe composition (solo)

160%

140%
2%
17%
120% 38%
100%
47%
80%

60% 18%
100%
40%
54%
20%

0%
Windstorm Flood Earthquake Hail Subsidence DiversificationTotal NatCAT

Page 88 of 153
© EIOPA 2011
 Graph 49: Manmade catastrophe composition (solo)

140%

120%
12% 17%
100% 2%

80%
64%
60%
100%
40%
9%
6%
20% 4%
19%
0%

Fire r ne it ility tion rrorism fication n-made

Moto Mari Cred Liab Avia Te rsi l ma
Dive Tota

The above graphs show the split between method 1 (scenario-based) and method 2
(factor-based) at EEA level, as well as the split between natural catastrophe and man-
made catastrophe. We can also see that at EEA level the principal components of
natural catastrophe risk were windstorm and earthquake, and that by far the largest
component of man-made catastrophe was liability, followed by fire and terrorism.

It should be noted that CAT risk, by its very nature, varies substantially between
different regions, in both its size and its composition – graph xSCR14 in the annex
show the breakdown of natural and man-made catastrophe risk charges by countries.
Nonetheless, feedback on the sub-module was quite unanimous, with only some
differences based on the characteristics of different countries.

This sub-module attracted a very large number of comments and complaints from the
non-life industry across Europe. The feedback can for the most part be classified into
four major areas: 1) calibration and methods used, 2) applicability to the respective
line of business or regional market, 3) data availability, and 4) effort needed to
calculate the required capital.

The applicability of the standardised scenarios was often questioned and it was
reported that for some (specialist) insurers the scenarios were not appropriate for
their business, resulting in both under-estimation and overestimation of risk for
different undertakings. Some undertakings suggested that use of personalised
scenarios or USPs might resolve this.

Methodology and calibration

Many concerns were raised around the man-made scenario methodology. The natural
catastrophe (NatCAT) approach was generally questioned in the areas of calibration
and data availability.

Page 89 of 153
© EIOPA 2011
A considerable concern was that the CRESTA factors for NatCAT did not adequately
reflect the actual risk undertaken by the respective insurance or reinsurance
undertaking. Only one country mentioned that zones are not matched with national
data records. In another country only a couple of the participating undertakings
applied the approach based on CRESTA zones. Confusion was also experienced in
some countries where there were no pre-defined natural catastrophe scenarios, as to
how the sub-module should be approached in their absence; in particular, when
undertakings couldn‟t use method 1 for NatCAT but could for man-made CAT risk,
they reported queries around how to use method 2. A few participants mentioned that
NatCAT scenarios are not applicable to the transport line of business.

Although feedback on the calibration of the catastrophe scenarios tended to vary

between countries, a few trends could be identified: several respondents reported that
the windstorm scenario was too severe (though another found it too weak), some
central European countries found the flood calibration too high and for example in one
country mentioned that measures taken since the most recent severe floods have not
been taken into account. The earthquake and hail perils were both reported to be too
highly calibrated by one country. One country presented an alternative calibration as
an annex to its report. Some countries questioned the appropriateness of the natural
perils defined for them. There were also comments that the module was over-
calibrated for P&I clubs and represented a 99.5% VaR over one year for the P&I
industry as a whole rather than single members.

Some participants also indicated that country-specific risk mitigation tools/effects

should be explicitly taken into account. Some suggested that the possible maximum
loss (PML) should be used as a volume measure instead of total sum insured.

One undertaking noted that the NatCAT risk factors are applied to all property
exposures, including geographical locations that are not exposed to certain types of
CAT risk, leading to the CAT risk under QIS5 being overstated for this class of
business.

A significant number of countries raised concerns that the CAT scenarios were not
suitable for credit insurance and surety business: almost all suggested that it was too
severe, although one country felt that it was probably too low for surety business.
Participants raised concerns that the scenario involving the failure of their three main
exposures was of a lesser probability than 1 in 200, suggesting that it failed to take
into account undertakings‟ active management of their large exposures, as well as the
credit rating of the insured. There was also feedback that the recession scenario was
double-counting with the tail of premium and reserve risk. Some undertakings lacked
sufficient years of observations with which to determine the failure rate and therefore
some used method 2 which for one country seemed to give a much lower result.
Undertakings in one country stated that it was not clear whether the credit scenario
was applicable only to credit insurance or to suretyship insurance as well. Moreover,
there were no method 2 parameters for suretyship business. One country suggested
that group exposure in SCR 9.146a) should be defined as exposure to single
financial/capital group as some undertakings assumed that it referred to a group of
policies in the same product.

Another area that triggered considerable feedback was liability insurance and its
treatment. A number of countries suggested that the CAT results for it were overly
severe. It was particularly highlighted that the requirements were not well specified,
for example there was no information on the number and magnitude of claims making
up the gross loss, making it difficult to adjust for reinsurance. The specified scenarios
Page 90 of 153
© EIOPA 2011
were also found not to be appropriate for all undertakings. The required sub-division
of premiums into the lines of business specified in the technical specifications was
considered not to be appropriate and there were concerns that the sub-module failed
to take into account contract limits.

One country stated that both methods provided unlikely results for medical
malpractice (defined as Third Party General Liability in the QIS5 Technical
Specifications) and that a more granular approach was needed. Another country
indicated that construction risk was not correctly captured as the volume measure,
total sum insured, did not represent the potential loss within the year, but the
potential loss when the construction is finalised.

There was also feedback from a number of countries that the marine scenario was not
suitable for their undertakings: for example, it was difficult for a small insurance
undertaking to have any insight into the costs associated with a ship‟s collision with an
oil-drilling platform and the costs of stemming the oil flow from burst pipes. Where
undertakings did not insure either oil tankers or cruise ships, as was often the case in
some countries, they were forced to use method 2 which had a very high impact on
SCR. One country fed back that they regarded the marine scenarios in general as
particularly severe. P&I clubs reported that there were similarities between the
standardised scenarios for marine business and their internally elaborated scenarios.

The motor sub-module was found by a few countries to be under-calibrated,

unrealistic or difficult to follow. The definition of input data was unclear, especially
vehicle years. One undertaking referred to an inconsistency in the motor CAT risk
scenario as motor property damage exposure is not required to be input for
windstorm whilst the factor method applies a charge of 175% premium.

For the fire sub-module, a number of undertakings (especially captives) in one

country used Option 2 for the fire scenario as they felt that Option 1 did not reflect the
maximum loss that could be made. In these cases, the maximum loss was the policy
limit and not the largest total insured value (“TIV”). A 100% loss was not believed to
be a likely scenario. One undertaking also encountered problems with separating
industrial from commercial business for Option 2 and with estimating the largest
concentration of buildings. One country noted that the fire peril risk description
concerns business interruption as well as property damage, but that the input value is
sum insured under the Fire and other damage line of business only (and not liability or
financial loss).There were suggestions that estimated maximum loss should be used
instead of total sum insured.

Undertakings from one country indicated that it was not clear whether the terrorism
scenario applied only to officially announced terror acts or to unproven events as well,
and whether undertakings needed to calculate the capital requirement for terrorism
even if this peril was excluded from all their insurance contracts. One country
expressed concerns about the need to make expert judgements for specific terrorism
and pirate scenarios, and another mentioned the difficulty of taking into account
contract specificities. The terrorism scenario was described by one country as
unrealistic.

One country mentioned practical difficulties in calculating the aviation scenario.

Another country encountered problems with data in this sub-module, while finding
that using method 2 gave a low capital requirement compared to the risk.

Page 91 of 153
© EIOPA 2011
One country mentioned that the highest risk/exposure should be chosen based on the
net of reinsurance value (not the gross of reinsurance value) as, if gross values were
to be taken as the input, insurance undertaking could buy facultative deep
reinsurance for the highest gross of reinsurance risk exposure, lowering the SCR
capital charge after taking into account risk mitigation even if the undertaking was still
exposed to a high risk net of reinsurance through other exposures.

In five countries, undertakings queried the lack of an allowance for a limitation on the
ceded risks when providing insured values for the purposes of CAT risk calculations.
Undertakings also commented that there was a certain amount of double-counting
within the man-made scenarios, as a number of scenarios resulted in policy limit
losses, which can obviously only be breached once. In some cases, undertakings
noted that they used the policy limits as the maximum loss rather than the underlying
exposure information.

For method 2, there was feedback that it was not risk-sensitive or was often very
penal. Criticism was expressed that the method was neither well-defined (e.g.
premium split) nor an adequate reflection of risk.

One country indicated that particularly in the miscellaneous line of business, the single
factor may prove to be inappropriate in many cases. Moreover some products in the
Miscellaneous line of business, such as Extended Warranty business, had
characteristics that were very similar to Assistance business but Assistance business
had no catastrophe charge in the standard formula.

It was also a concern that there was no allowance for geographical diversification in
this method. It was suggested that the severity of method 2 created difficulties for
insurers with non-EEA exposures where method 1 could not be used. The large
discrepancy between the method 1 and method 2 results was also a concern raised by
a number of countries.

One country felt it was not clearly stated in the technical specifications whether the
method 2 factors should be applied to all premiums or just to those linked to
catastrophe risk. Undertakings in a few countries suggested that the use of total
premium for more than one peril was double-counting, and some proposed that the
premium should instead be allocated between the different perils (although others
indicated that if this was required, there were likely to be problems with data
availability).

It was suggested by participants that insurers could calculate their own gross
aggregate exposure at the 1 in 200 level (as part of their risk management system)
and use this information in the standard formula for catastrophe risk. To validate
insurers‟ work, supervisors could compare insurers‟ estimates against a measure of
exposure and investigate outliers.

One country indicated that the CAT risk sub-module did not fit workers‟ compensation
insurance.

Finally, some undertakings suggested that there is an overlap between the CAT risk
capital requirement and binary events already taken into account in technical
provisions.

Page 92 of 153
© EIOPA 2011
Data availability

Almost all countries observed problems with data availability to a greater or lesser
extent, especially in the man-made scenarios for example number of insured buildings
in a 150/300 metre radius, availability of postal code for the insured object (e.g.
buildings under construction, large cranes or technical equipment), for those
insurance contracts where the whole production of a given calendar year is insured,
and for the terrorism sub-module. This forced participants to use the factor-based
method 2, whose calibration was found to be excessively prudent. Two countries
indicated problems with allocation to zones for multi-location policies and one country
with segmentation of individual contracts into CRESTA zones for group policies. In
some countries, total insured value (“TIV”) was not available for all CRESTA zones,
especially for small undertakings. In other cases, the TIV was not available by CRESTA
zone, only in total. Two countries mentioned difficulties with reliably removing
catastrophe losses from historic data creating a significant risk of double-counting
with the premium and reserve risk sub-module. In some cases participants suggested
using probable maximum loss instead of total insured value.

Reinsurance issues

A number of countries also commented that the sub-module was not appropriate for
reinsurers, especially the man-made scenarios. The participants had problems with
identifying the precise location of all large risks and the share of total sum insured.
Usually the coinsurance structure of large risks is not known, and the reinsurance
client does not provide limit profiles on an original total sum insured basis.
Furthermore, in contrast to NatCAT perils, a standardised concept for geo-coding large
risks potentially exposed to man-made CAT scenarios does not exist. Additional
information on loss history, limit profiles relating to the reinsured portfolios, controls
on treaty capacity and expert judgment on the accumulation potential of reinsured
covers had to be taken into account. In addition, reinsurers complained that the
failure to split total sum insured values per zone into the underlying lines of business
produced dramatic mis-estimation of the NatCAT loss potential. According to one
country, the proposed set of factors represented an “average cost approach per zone”
irrespective of the regional characteristics of the set of large NatCAT events required
to define the 1 in 200 event, which would be relevant to capture the geographic
spread of a specific NatCAT portfolio. Cover-specific loss caps (e.g. event limits,
annual aggregates) could not be taken into account under a format which aggregated
clients‟ exposure data as a first step. Furthermore, loss-reducing elements (e.g.
deductibles applicable at original policy level) could not be taken into account
appropriately. The calculation of the risk-mitigating effect of reinsurance business
corresponding to intra-group operations was not possible as retrocession protection
covered third party contracts as well. For reinsurance portfolios combining Motor Own
Damage and Motor Third Party Liability covers it was difficult to determine the number
of vehicles.

Some supervisors remarked that for reinsurers, the complex nature of their business
should be addressed by partial internal models.

Reinsurers‟ feedback on their evaluation of the results derived by the different

methods was split. Some participants stated that the results derived by the factor-
based method were much higher than their partial internal model. However, one
participant considered that the factor-based method understated the NatCAT scenario
loss potential. It was commented that the factor to be used depends heavily on the
NatCAT related premium level to be taken as the calculation basis.
Page 93 of 153
© EIOPA 2011
Lack of clarity

Participants also asked for more guidance in the following areas: how to treat
company-specific deductibles in NatCAT, the procedure for calculating the
reinstatement premium (since such reinstatement will need to be defined taking into
consideration the undertaking‟s remaining risk exposures as from the forecasted
event; it needs to be clarified when in the year the forecasted event takes place),
treatment of products that cover several risks, definition of volume measures used for
calculating CAT risk, definition of exposure (turnover insured or exposure, outstanding
or total, and at which point in time), and also the influence of the choice of the
different x-year scenarios (calculations in the tool stipulated 5-year and 10-year
scenarios).

General remarks

Several countries indicated that the industry generally considered the CAT sub-module
to be too complex, inappropriate or difficult to implement, requires too much data and
does not properly capture the risk. Two countries indicated that judgment was needed
to decide which CAT scenarios applied, which led to a lack of comparability between
results.

A few supervisors were also of the opinion that the CAT sub-module should be
simplified (for example by reducing the number of scenarios) and in some cases felt
that the calibration should be changed to avoid penalizing specialised non-life insurers
particularly impacted by the high calibration of CAT risk. Two supervisors and
undertakings in two other countries indicated that the treatment had not been
appropriate for some niche players (for example there was no man-made CAT for
legal expenses insurers or assistance insurance undertakings).

There was also a statement from one country (supported by the supervisor) that the
combined non-life premium and reserve risk and CAT risk capital charges should not
exceed aggregate limits in place in treaties accepted and retroceded. Two countries
felt that the capital requirement for CAT risk is an inappropriately large component of
the SCR.

Most undertakings had lower partial internal model capital requirements for CAT risk
than those calculated according to the standard formula, but there were also some
with higher or similar results.

5.10. Undertaking-specific parameters

5.10.1. Participation in the USP part of the exercise and comparison of

USPs with standard parameters
It must be underlined than for the vast majority of countries, the participation was
negligible with no more than five undertakings in any given line of business
responding to this section of the exercise. In five countries, the sample size was
limited to at most six to ten participants per line of business and the sample was of
sufficient size for analysis only for the most popular lines of business in three big
countries.

Page 94 of 153
© EIOPA 2011
For lines of business like workers‟ compensation, non-proportional health, casualty
and MAT reinsurance, there were no more than eight users in the whole of Europe.
The most popular areas for the calculation of USPs were fire and other property
damage (about 100 participants in the whole of Europe), motor vehicle liability, third
party liability and motor other classes. Lines of business like medical expenses,
income protection, MAT (marine, aviation and transport), legal expenses and
miscellaneous also attracted some participants using USPs (46-76 participants).

Therefore the sample can not be regarded as representative. The main reasons given
for failure to participate in this part of the survey were a lack of time (since the
calculation was optional efforts were concentrated on other issues) and a lack of data
consistent with the Solvency II format.

The median USP amounts for individual lines of business are presented in the tables
below (some lines of business have been omitted as the sample is too small to draw
any conclusions or calculate the median).

Table 14: Premium risk USPs

LoB* Standard parameter Median of USP Sample
size
Health - medical expenses 4% 4.1% 77
Health – income protection 8.5% 7.3% 76
Non life – motor vehicle liability 10% 7.7% 106
Non life – motor other classes 7% 6.8% 99
Non life – MAT 17% 13% 60
Non life – fire 10% 8.4% 116
Non life – third party liability 15% 10.7% 105
Non life – credit 21.5% 20.0% 30
Non life – legal expenses 6.5% 4.9% 46
Non life – assistance 5% 6.0% 22
Non life – miscellaneous 13% 9.8% 40
* Sample size for workers‟ compensation, non-proportional health reinsurance and non-life non-proportional
reinsurance (property, casualty, MAT) too small to be included

Table 15: Reserve risk USPs

LoB* Standard parameter Median of USP Sample
size
Health - medical expenses 10% 11.6% 59
Health – income protection 14% 12.0% 61
Non life – motor vehicle liability 9.5% 7.4% 89
Non life – motor other classes 10% 10.2% 75
Non life – MAT 14% 13.3% 44
Non life – fire 11% 10% 87
Non life – third party liability 11% 8.4% 86
Non life – credit 19% 18.9% 26
Non life – legal expenses 9% 6.5% 34
Non life – assistance 11% 12.4% 14
Non life – miscellaneous 15% 18.2% 22
* Sample size for workers‟ compensation, non-proportional health reinsurance, non-life non-proportional reinsurance
(property, casualty, MAT) too small to be included

The USPs were in most cases lower than the standard parameters. However, in some
lines of business a significant standard deviation was observed (in both premium and
reserve risk for MAT, credit, miscellaneous and non-proportional property reinsurance,
and in reserve risk only for health medical expenses, and health income protection).

5.10.2. General comments

Page 95 of 153
© EIOPA 2011
Participants commented variously that the use of USPs could represent a barrier to
entry as new undertakings would not have the relevant and requested data, that it
was a way to lower their capital requirements and that small and medium-sized
insurers will not be capable of generating the statistical data base needed.

Some undertakings proposed allowing entity-specific lines of business with specific

parameters and calibrationor the replacement of all standard formula parameters with
USPs, and many undertakings in different countries proposed extending the
application of USPs to various different modules (see below). But there were also
comments that the standard formula is already too complex and that the use of USPs
should not make the standard formula resemble internal models and therefore the
number of USPs allowed should not increase. Additionally USPs should not function as
“internal modelling lite”, that is, a way to reduce the SCR without the rigorous
qualitative requirements of internal modelling.

In general industry considered the requirement to have at least fifteen years‟ data
available too high a hurdle (in fact according to the technical specifications there is
weighting with market parameters so that the minimum requirement is five years)
and that the requirement for five years‟ data would penalise SMEs and recent start-
ups.

Some undertakings in one country commented that it seemed difficult to justify

variations in parameters from undertaking to undertaking in what is supposed to be a
standard method: undertakings that believe that different assumptions would be
appropriate to them should apply this within an internal model and go through the
associated approval process. In this way, the relevance of all the assumptions can be
considered together as a package and there will be no „cherry picking‟ of particular
parameters they would like to change.

Some undertakings from several countries mentioned that they would prefer country–
specific parameters.

Many undertakings from one country commented that there is no optimal approach to
determining USPs, as all approaches have pros and cons. As a consequence, they
proposed allowing further alternative approaches not already captured in the technical
specifications. For instance, undertakings wanted to be allowed to choose methods
adapted to each line of business instead of using the same standardised approach for
all lines of business. These comments appear to be a misunderstanding of the
requirements, as according to the technical specifications undertakings can use
different methods for different lines of business.

For some undertakings, it was not clear from the technical specifications how they
should decide if it would be appropriate to use a USP in place of a standard factor and
they asked for more clarity on the conditions of application of the standardised
methods, on which of the USP methods should be used under which circumstances,
and on how this could be assessed. Undertakings would appreciate more detailed
guidance and description of the standardised USP methods, as well as instructions for
preparing data.

Undertakings in one country raised the issue of the calculation of USPs at points in
time other than year end, as may be required for quarterly reporting or in case the
SCR has to be recalculated: they argued that due to limited availability of input data,
a projection or estimation of the required data would be necessary.

Page 96 of 153
© EIOPA 2011
5.10.3. Assessment of QIS5 methodology for USP

One country commented that the standardised methods did not cause any major
concerns.

The following general discrepancies with the QIS5 USP methodology were mentioned
by participants:
In cases where data availability was variable, the three methods led to very
different (sometimes contradictory) results.
There was a problem with the meaningful estimation of risk if the volume
measure was small or the business was rapidly growing.
Methods were not applicable for reinsurance business.
Methods were too specific and too restrictive to be used to determine sensible
USPs adequately reflecting the risk faced by the undertaking.

The most frequently mentioned problems in USP calculation for undertakings -referred
to by thirteen countries - related to a lack of appropriate data (not compatible with
the QIS5 requirements, collected under different systems, for example under different
segmentations, in too short time series, unavailability of net best estimate ultimate
after one year, lack of Solvency II-based figures such as historic best estimates, no
data for revision risk). Other countries did not respond to this question as their
undertakings did not test the USP calculation.

One country mentioned a lack of time for testing the proposed methodology as the
main difficulty.

Regarding the methods tested by undertakings for premium risk, the most popular
methods were the first one (eleven countries) and the second one (seven countries),
which were based on the variation of ultimate loss. The fulfilment of assumptions was
given as the reason for adopting these methods and some undertakings mentioned
that the results were similar. Method 3 was rarely used due to higher results or higher
data standards (three countries mentioned problems with data in this context).

For reserve risk, undertakings mostly used method 2 (nine countries) and method 3
(six countries), based on the Merz/Wüthrich approach. Countries mentioned as
justification for using these methods: comparison with the Mack standard deviation,
the theoretical basis (for method 2), and less demanding data requirements, since
paid claim triangles are straightforward to extract. However, undertakings also
commented that these two methods could not cope with zero or negative numbers in
the first development period, and did not allow for model error. Some undertakings
remarked that the results produced by methods 2 and 3 were low for some large
motor portfolios or seemed unreliable. Regarding method 1, some participants found
that it produced results consistent with their own internal analyses. The negative
comments on method 1 were that the one-year reserve movement required a
significant amount of time for the extraction of data, or that results were not valid.

In two countries all of the methods were tested by at least some participants.

Two undertakings used all relevant methods and calculated the average value. Two
countries mentioned that undertakings calculated USPs using their own methods, so
the results were not comparable.

Page 97 of 153
© EIOPA 2011
One participant stated that methods 2 and 3 did not sufficiently take into account the
premium margin/business cycle risk.

5.10.4. Suggested changes and improvements

Participants were asked to provide their own suggestions for other parameters which
could be replaced by undertaking–specific ones. Many answers were outside of the
scope provided by the Directive, for example requesting USPs in the following areas:
counterparty default risk, market risk (especially property, equity), operational risk,
correlation matrices at country level, CAT risk scenarios.

The proposals consistent with the Directive were as follows: parameters in the non-life
CAT sub-module (especially in factor-based methods and for credit and suretyship),
parameters in life CAT, parameters in expense risk, parameters in biometric risks
(longevity and mortality risks), parameters in lapse risk and parameters in health for
one country-specific system.

Participants suggested the following methods:

• Mack‟s formula and methods developed by Merz and Wüthrich, especially the
bootstrap version of the MW method, the stochastic version of the Bornhuetter-
Ferguson method (a standard in reserving practice), the ODP Bootstrap which is an
industry standard method or more generally bootstrap on GLM and bootstrap
techniques in general adjusted to a one-year time horizon, stochastic reserving
methods in general, methods based on the standard error estimated according to the
models used for determining the best estimate of technical provisions, and net
casualty ratios.
• For premium risk, enhancing the current methods by allowing for the effects of
premium/underwriting cycles or the specific characteristics of individual lines of
business.
• An AR2 process to remove volatility inherent in underwriting cycle.
• A method based on frequency/severity widely used by insurance sector.

Some undertakings in five countries suggested using individual NatCAT models, either
modelled internally or from external sources (e.g. RMS, Willis, AON Benfield, etc.).

5.10.5. Source of data, adjustments, assumptions, difficulties with

data

Fifteen countries answered that in most cases internal data were used (claim
triangles, historic earned premiums and incurred claim costs). Some countries also
mentioned external sources (the previous insurer).

The calculation required various data adjustments and assumptions, for example:
splitting by lines of business on the basis of expert judgment as the right
segmentation was not available;
stripping out large catastrophe losses;
ultimate losses at the end of 2009 and not as recommended from the end of
each accident year;
undiscounted best estimate;
underwriting year as accident year was not available;
interest roll-up for discounted reserves; and

Page 98 of 153
© EIOPA 2011
 calculating best estimate for earlier year claims using current regression
coefficients and applying them to previous years' history under the assumption
that current development is not different from previous years.
No mention was made of adjustment for inflation.

The most frequently mentioned difficulties were as follows: lack of data in the right
format (available data were collected under the Solvency I regime), problems with
different segmentations, isolation of large catastrophe claims and availability of
historic net data.

5.10.6. Inflation adjustment

In many countries most undertakings were of the opinion that inflation is

appropriately reflected in the data and that past experience is representative for the
future, or even took the view that inflation has no material effect and so no
adjustment is necessary.

Some participants pointed out that in practice such an adjustment would be

challenging, since the impact of inflation would depend on a multitude of factors and
would be likely to vary across different lines of business, different countries and
different currencies. It would also be impacted by insurer-specific conditions (for
example in relation to policy limits or contractual agreements), which would make it
difficult to compare estimates across different insurers. Undertakings in one country
were of the opinion that reliable estimates of the historic impact of inflation on claims
are not readily available and that it can be difficult to strip out historic claims inflation,
particularly for some lines of business (e.g. those with significant elements of
commercial insurance). According to one country it is not possible to adjust for
hyperinflation.

Undertakings from one country remarked that premiums should be adjusted

correspondingly, as otherwise there would be inconsistency in the calculations.

In some countries a few undertakings considered inflation adjustments.

5.10.7. Supervisory views on USPs

One supervisor mentioned that undertakings had realised a considerable reduction in

risk capital for non-life through the calculation of USPs and suggested developing
USPs for reinsurance as many problems had been encountered in adequately taking
account of risk mitigation techniques both in calculating the adjustment factor for
premium and reserve risk and in the catastrophe scenarios.

In the opinion of another supervisor, USPs should be permitted for mortality and
longevity as long as the insured group is large enough. Another supervisor indicated
that the scope of application of USPs should be enlarged to cover almost all
underwriting risk factors as an alternative for small undertakings. Additionally the
calculation methods should be less prescriptive and more principles-based.

There were, however, some supervisors who argued against enlarging the number of
USPs to prevent any evolution of the standard formula into internal modelling and to
avoid cherry picking. One supervisor highlighted the need for balance in this context
and mentioned the burden USPs pose for supervisors.
Page 99 of 153
© EIOPA 2011
Three supervisors were of the opinion that the low number of undertakings adopting
USPs was caused by lack of time and/or data, or the strict requirements. One
supervisor noted that undertakings only had limited historical data, so the proposed
use of internal data was unrealistic. One supervisor strongly opposed any softening of
data quality standards by, for instance allowing reference to Solvency I technical
provisions when determining USPs.

A few supervisors acknowledged there were issues with the appropriateness of the
current catastrophe calculation, but were of the opinion that natural catastrophe
factors could not be undertaking-specific due to lack of relevant experience and the
short data series available.

One supervisor mentioned that a significant number of undertakings had used USPs
and had not highlighted any particular difficulties.

Regarding methods, one supervisor expected the non-life calibration exercise to

produce some other methods for calculating USPs; another agreed with its industry‟s
opinion that the USP for revision risk should be reviewed.

It was indicated that additional guidelines and conditions of application were

necessary. One supervisor would welcome a more prescriptive description than in the
technical specifications regarding the ultimate after one year in method 1 for standard
deviation for premium risk.

One supervisor was disappointed that relatively few participants calculated USPs:
undertakings seemed to have difficulties with the data requirements for historic
technical provisions on a Solvency II basis.

5.11. Risk mitigation

Risk mitigation techniques other than proportional reinsurance were generally seen as
difficult to take into account within the standard formula, and a considerable number
of participants reported problems relating to this topic. Concerns were mainly raised in
the context of the non-life underwriting module. However there were also second-
order effects extending to the counterparty default module, because there the risk-
mitigating effect of reinsurance arrangements has to be taken into account with
reference to its impact on the risk charges for other modules.

5.11.1. Non-proportional reinsurance adjustment in non-life

Most undertakings failed to determine the adjustment for non-proportional

reinsurance in the premium risk factors because of problems with data availability.
The calculations were also seen as too complex.

Table 16: Non-proportional reinsurance adjustment in non-life

Line of business* Median of non-proportional reinsurance Sample size
adjustment (EEA)
Non life – Motor vehicle liability 81.2% 87
Non life – Motor other classes 99.4% 39
Non life – MAT 77.9% 27
Non life – Fire 81.4% 81
Non life – Third party liability 81.7% 87
* Sample size for credit, legal expenses, assistance and miscellaneous too small to be included

Page 100 of 153

© EIOPA 2011
In fourteen countries no undertakings calculated non-proportional reinsurance
adjustments. In ten countries the sample was at most seven undertakings in any line
of business. The biggest sample for a single line of business was from one of the
larger countries; it comprised 23 undertakings (in the most popular line of business)
and was regarded by the supervisor as “sizeable”.

A considerable number of countries raised concerns that the method for determining
the non-proportional reinsurance adjustments in the premium risk factors was not
suitable for every kind of non-proportional reinsurance (for example whenever the line
of business was covered by more than one excess of loss treaty or in the case of the
Marine, Aviation and Transport line of business).

Two countries mentioned that use of the formula had potentially undesirable
consequences. For large undertakings with high retentions the adjustment would often
be close to unity indicating no significant adjustment to the factors. For small
undertakings the adjustment would often be materially less than one. The overall
premium and reserve risk would therefore be smaller for small undertakings than
large ones, which seems contrary to expected outcomes. There was also feedback
that the definition of the data requested for the calculations concerning the
reinsurance treaty was not entirely clear; in particular it was not obvious how to take
reinstatements into account. Furthermore it was reported that the risk-mitigating
effects of facultative reinsurance could not be taken into account, which often had a
significant impact on the capital charge. One undertaking queried whether the
average cost of claims needed to be adjusted for inflation, and if so, what index was
specified.

Other difficulties concerned the requirements for detailed historical claims data which
was not always available (for example in the case of claims in long-tailed lines of
business which have not been fully settled, or in constructing a reinsurance claims
triangle). It was commented that the calculation of netting down factors was also
relatively complex and that changes to a reinsurance programme over time made it
hard to obtain historic claims data net of the current reinsurance programme. Some
undertakings encountered difficulties in calculating the duration of recoverables, and
assumed the duration of the recoverables to be the same as the duration of the claims
outstanding provisions.

Reinsurers commented that for a reinsurance portfolio a realistic derivation of the

adjustment factor for non-proportional retrocession would require more detailed
partial internal modelling. One undertaking also expressed a view that the USP
approach fitted better and ensured a homogeneous treatment between lines of
business.

One supervisor expressed the view that despite its limitations, this adjustment could
be maintained in the Level 2 draft text. In thirteen country reports this issue was
mentioned as one of the most important discrepancies in the non-life underwriting risk
module. One country remarked that the adjustment should be considered further but
that the tested methodology is inappropriate, and two others suggested the issue
could be resolved with partial internal models. In one country several undertakings
felt that they did not have sufficient time to gather the data and perform the
calculations for the QIS5 exercise, but that they would be able to in due course.

Some undertakings reported the following adjustments and assumptions made in

order to perform the calculation:
Page 101 of 153
© EIOPA 2011
 netting down for non-life reinsurance and using available large claims data;
using multiple years of data for the classes of business (general liability and
motor liability) to which their non-proportional reinsurance applied, as those
lines of business could have volatile results;
using loss ratio assumptions that did not include any very large risks that would
be reinsured with non-proportional reinsurance;
using the approximations in the helper tabs for average outward attachments
and limits; and
applying the gross average claim cost to all claims, with no capping or limit.

Undertakings made some comments about changes they considered could be made to
the adjustment for non-proportional reinsurance in the premium risk factors. The
suggestions were as follows:
Reinsurance calculations for premium and reserve risk should be extended to
include other non-proportional reinsurance structures (e.g. stop loss).
Capital charges for premium risk and reserve risk should be split, and allowance
should be made for stop loss cover in reserve risk too.
Using the observed volatility of the gross and net loss ratios to estimate the
impact of the reinsurance on volatility.
A distinction between mass claims and large claims, although it was noted that
large claims have already been partly covered by the CAT sub-module.
Introduction of a Pareto-based model in addition to a lognormal one, to more
precisely model the skewed distribution.
Deduction of the part of the stop loss cover that would be used for the segment-
specific 1 in 200 year event (premium risk).
Calculation of adjustment factors for all sub-segments and use of the premium
weighted average as the overall adjustment factor (any segment that is not
covered would have an adjustment factor of 1).
Market data on average claims and standard deviations should be made available.
Make an addition for captives as the loadings are too harsh.
Use only large losses above a certain threshold, so less data is needed to
calculate the adjustment.
More flexibility to allow for possible retention clauses (e.g. aggregate
deductibles).
To allow for risk mitigation in the same way as in catastrophe risk.
For EIOPA to develop an industry standard increased limit factor curve for each
class.

5.11.2. Risk mitigation techniques in non-life CAT

In the QIS5 Technical Specifications, participants were asked to take into account
their own reinsurance programmes. Many countries mentioned that in the catastrophe
risk sub-module the effect of risk-mitigation was difficult to take into consideration in
the standard calculation (especially in the man-made scenarios) and that its impact
could therefore be underestimated.

Some countries referred to more specific difficulties which undertakings faced with risk
mitigation, such as: including stop loss contracts (two countries) or excess of loss
treaties (two countries), factoring in non-proportional reinsurance, netting down in
case of facultative contracts (four countries), dealing with cases of more than one
treaty on a line of business, netting down for quota share reinsurance with
reinstatements. Some problems arose from the interaction between modules and the
application of a reinsurance treaty to a number of catastrophes separately (without

Page 102 of 153

© EIOPA 2011
needing to reinstate, or as many times as there are reinstatements). There was also
uncertainty, especially in liability insurance, about what the nature of the gross loss
was (e.g. large single claim or multiple smaller claims) and thus how reinsurance
programmes should be applied, as this impacts on reinsurance recoveries.

For a natural catastrophe reinsurance portfolio including both pro rata and non-
proportional covers the loss cannot simply be added up at a per-peril level. It was
difficult for undertakings to merge approaches appropriately.

One country was of the opinion that the scenario approach was difficult to implement
when the scenario impact exceeded the limits of a reinsurance treaty.

Related concerns were raised by reinsurers, who indicated that the standard formula
is not able to properly take the complexity of their business model into account (see
non-life catastrophe section above).

5.12. Participations
An amount of €377bn was reported in the QIS5 balance sheet in respect of
participations. See section 3.4 for details on the valuation methods employed, and
section 8.5.3 for details of the adjustment to basic own funds for participations in
financial and credit institutions.

Under QIS5 participations in related undertakings were subject to a 22% risk charge
where the participation was considered strategic; otherwise the appropriate global or
other risk charges of 30% or 40% respectively were to be applied. One country felt
that the differentiation between strategic participations and ordinary equity
investment was dubious and could lead to regulatory arbitrage.

The table below analyses participations over the different categories, as a share of
both the total number (1034) and total value (€1094bn14) of participations. It shows
that overall undertakings regarded two thirds of their participations as strategic,
applying a capital charge of 22%. Nearly all participations in insurance undertakings
were considered to be of a strategic nature. Other types of participation were
considered strategic more often than not. Where participations were not considered
strategic and were therefore subject to a standard equity risk charge, the „global
equity charge‟ was generally applied (30% risk charge) instead of the „other equity
charge‟ (40% risk charge).

Table 17: Equity charges on participations

Equity Share of total Share of total
charge number value of
participations participations

Participations in financial and credit N/A 13% 5%

institutions
Participations excluded 100% 1% 0%

Insurance strategic participations 22% 25% 30%

14
This figure does not tally with the balance sheet figure given previously (€377bn) as not all of the participation
categories considered here were classed as „participations‟ for balance sheet purposes (such as „strategic investment
other‟).
Page 103 of 153
© EIOPA 2011
 Insurance non-strategic participations 30/40% 3% 2%
Strategic investment other 22% 36% 38%

Other related undertakings - global 30% 8% 25%

equity
Other related undertakings - other 40% 13% 1%
equity

5.12.1. Strategic participations

The specifications did not describe the criteria to be applied in determining whether a
participation was strategic or not. However the qualitative questionnaire asked
undertakings to describe the criteria they had used in distinguishing strategic
participations from other participations.

Responses to this question were provided by 28 out of 30 countries, although some

noted that their response was based on data from only a few participants.

The criteria adopted are summarised in the following table. This sets out the range of
approaches reported. In some cases these were freestanding but often they were
combined. The third column indicates the most frequent combinations that were seen.

Table 18: Criteria for determining strategic participations

Number Criterion Number of Combined with
countries other criteria
reporting
1 Long-term nature of the relationship 15 5, 4
2. Participation is controlled (>50%) or 15 5,1,4, 7
wholly owned and/or fully consolidated
3. Based on holding >20% 11 5,1,4,7
4. Long-term involvement in operations, 8 1, 2/3
management or board
5. Participation is intended to maintain or 20 1, 2/3
develop the activities of the participating
undertaking or support its business
model
6. Insurance participation or core to 8
insurance activity
7. Ancillary or support to the participating 7 2/3
undertaking (e.g. sales, IT, premises,
staff, administration of investments)
8. All participation except where disposal 5
decided, likely or possible
9. All participations except those made as 3
part of investment strategy
10. Driven by treatments as associates 2
under accounting
11. Participation not in run-off 1
12. Participating undertaking holds a 1
blocking minority
13. Participation represents > 1% of assets 1

14. All participations 2

From this it can be seen that the emphasis was most frequently on a combination of:
the degree of control – often citing control/full consolidation or subsidiary
status;
the long-term nature of the relationship or involvement in the participation; and

Page 104 of 153

© EIOPA 2011
 the maintenance or development of the activities of the participating
undertaking or support to its business model.

Items 6 and 7 can be regarded as a more granular expression of the same concept as
the third bullet above.

It is notable that in only 2 countries was the view that all participations are strategic
explicitly stated in the range of responses by industry.

Page 105 of 153

© EIOPA 2011
6. SCR – Internal model
In the Solvency II regime the Solvency Capital Requirement (SCR) is to be calculated
by undertakings in accordance with the standard formula (discussed in the preceding
chapter) or using a full internal model, or using a combination of both a partial
internal model and the standard formula. For QIS5, undertakings that are developing
full or partial internal models were asked to calculate the SCR both with the standard
formula and with the internal model. Additionally participants were asked to provide
quantitative data in order to allow the impact of the use of internal models on solo
undertakings‟ and groups‟ capital requirements to be assessed. In order to collect
information on the current and future potential use of internal models in the EEA,
there was also a qualitative questionnaire directed to all undertakings.

It should be emphasised that any conclusions drawn from the information on internal
models are only representative with respect to the sample of responses provided,
which in some cases was very small. Using the results from this small sample to infer
anything about the general EEA-wide population might lead to biased conclusions, and
hence the observations on internal models should be interpreted with caution.

Disclaimer: Due to the fact that undertakings‟ internal models have not yet been
finalised and because of the small sample provided, no exact conclusions can be
drawn as to the size of the capital requirements calculated by internal models
compared to the capital requirements calculated by the standard formula.
Furthermore some undertakings mentioned using internal model techniques which in
EIOPA‟s opinion were not in accordance with the Level 1 text and the QIS5 Technical
Specifications.

6.1. Internal models on solo level

The qualitative questionnaire for internal models was to be completed by
undertakings:
which are not part of a group and which currently use or intend to use an
internal model;
which are part of a group and are intending to use an internal model other than
a group internal model for the solo SCR calculation; or
which are part of a group and are intending to use a group internal model to
calculate the solo SCR.

From the qualitative questionnaires it was found that:

262 undertakings (out of 309 which answered the question) were already using
internal models for some individual aspects of their business; whereas
289 undertakings were currently working on the implementation of their
internal model for Solvency II purposes.

Solo undertakings which were part of groups for the most part declared that they
would be using internal models developed at group level; 159 out of 166 (96%)
undertakings answered that they used the same methodology as the one used in the
internal model for the calculation of the group SCR. Not many undertakings which
belonged to groups reported that there were assumptions used for the calculation of
the group SCR which did not fit their risk profile.

Most solo participants that submitted answers to the qualitative questionnaires

referred to group internal models which would be used for both group and solo SCR
Page 106 of 153
© EIOPA 2011
calculation. Local undertakings that are currently implementing internal models (in
many cases they are only in the early stages) presented very detailed group answers
without giving much information on the local level. This meant that EIOPA could not
conduct any detailed analysis of the specificities of the solo calculation compared to
the general characteristics of the internal model.

Below are presented the most common solo undertakings‟ comments about deviations
from group internal models:
solo undertakings used local information for calibration of the internal model
(mentioned by ten countries);
correlations among the various non-life lines of business were different at local
level (mentioned by five countries);
operational risk might be calculated by the standard formula at solo level
(mentioned by three countries);
the local internal model will use a different methodology for catastrophe risk
(mentioned by one country); and
the diversification factor will be undertaking-specific (mentioned by one
country).
A few undertakings reported that they would not look to develop a specific internal
model because the benefits of developing such an internal model would not outweigh
the costs.

Regarding the cooperation between groups and local entities, almost all groups
declared that they feed entities with the data and discuss with them the methodology
and especially the local specificities. Some of the groups only validate the solo results,
whereas others carry out the calculation at the ultimate group level.

It was noted that in all EEA countries which provided internal model results, many
undertakings intend to submit applications just before or just after the introduction of
the Solvency II regime. Some undertakings (which belonged to groups) commented
that they have already been involved in the pre-application process because of their
parent undertakings and also local supervisors.

It is worth mentioning that several supervisors reported that many undertakings

indicated that they were going to use internal models to calculate SCR under the
Solvency II regime (in many cases they had already entered into the pre-application
phase) but did not submit any QIS5 results (either qualitative or quantitative)
regarding internal models.

6.2. Current status of internal modelling in the EEA

Participants reported the following main reasons for using internal models instead of
the standard formula:
internal models better reflect the undertakings‟ specific risk profiles, additional
risks are covered by the internal model beyond those covered by the standard
formula;
the internal model applies a more granular aggregation method than the
standard formula;
the standard formula does not take into account volatility; and
to use IFRS valuation rules instead of QIS5.
In relation to the last comment, we note that internal models may be used to
calculate the SCR but should not change the approach to valuation.

Page 107 of 153

© EIOPA 2011
Undertakings also mentioned using different (internal) parameters to the standard
formula in order to take into account the specific risk profile of the undertaking. But
according to article 104 of the Level 1 text and the QIS5 Technical Specifications there
is a restricted and closely-defined area where undertaking-specific parameters can be
used. EIOPA‟s view is that changing the parameters of the standard formula
themselves should not be considered as internal modelling and does not comply with
the Solvency II requirements regarding internal models.

One supervisor mentioned that natural catastrophe risk might also be an area where
internal models might better capture the undertakings‟ risk profiles than the standard
formula. However another supervisor noted that even though CAT risk may be very
substantial, particularly for non-life insurers, undertaking-specific partial internal
models are often not a practical solution given the small size and relatively large
number of non-life insurers. Another supervisor reported that monoline insurers
claimed that the standard formula was not suitable for their business.

Regarding the structure of the internal models, some undertakings reported adopting
a modular approach similar to the standard formula, some undertakings reported an
approach broadly similar and some used approaches completely different to the
standard formula.

Some individual undertakings also made comments on the parameters used in their
correlation matrices. The internal models‟ correlation parameters varied from the
standard formula in most cases from ±25% to ±50%. Some examples of differences
between the standard formula correlation parameters and internal model ones are
given below:
between operational risk and BSCR: 50% or 75% (standard formula 100%);
between counterparty default risk and market risk: 50% (standard formula
25%);
between non-life underwriting risk and counterparty default risk: 25%
(standard formula 50%);
between health underwriting risk and life underwriting risk: 50% (standard
formula 25%);
between interest risk and property risk (up): -50% (standard formula 0%);
between interest risk and currency risk (up): -50% (standard formula 25%);
between concentration risk and equity risk (up and down): 75% (standard
formula 0%); and
between spread risk and concentration risk (up and down): 75% (standard
formula 0%).

6.2.1. Internal Model changes

Individual undertakings were at various stages in the development of their policy on

internal model changes. Many undertakings are planning or currently working on the
internal model change policy. Other undertakings reported that the process of defining
major and minor changes was still under development.

Criteria which could be applied to distinguish between major and minor changes, as
reported by undertakings, were the following:
impact on capital requirements, for example using a threshold of change in the
SCR - this criterion was mentioned by almost all countries;
changes to the methodologies used to perform calculations;
changes to the structure of the internal model;
changes in the assumptions/parameters of the internal model;
Page 108 of 153
© EIOPA 2011
 changes in modelling strategy, for example the aggregation technique;
changes in governance;
changes affecting the Solvency II compliance of the internal model; and
changes in the underlying business.

Groups also mentioned using the outcome of sensitivity analysis before and after the
change as a criterion, as well as whether or not there was a significant change in
exposures.

Some undertakings also reported using internal committees to ensure the

effectiveness of the internal model changes; for example, some undertakings reported
substantial changes to the board in order to receive sign-off from the relevant board
committees.

One supervisor mentioned that the borderline between major and minor changes was
a difficult question and that both qualitative and quantitative thresholds might be
needed. Another supervisor highlighted that in some cases while the process of the
internal model changes was well organised, the definition of a minor or major change
was still lacking.

6.2.2. External models

There was a wide range of responses that were provided about the external models
likely to be used by undertakings. Participants which took part in the QIS5 study
mainly used external models/programmes in the following areas:
Natural catastrophe risk models.
Economic Scenarios Generators.
Tools for the calculation of the best estimate.

Undertakings‟ answers show that they mainly use external models/programmes to

calculate catastrophe risk and to perform stochastic and actuarial simulation of cash
flows.

One of the main concerns regarding external models, in the opinion of one supervisor,
regards potential black box issues and the risk that some undertakings might not
meet the documentation requirements as a result. Another supervisor mentioned that
the real concern for them has arisen from models built or sold by vendors/consultants.

6.2.3. Probability distribution forecast

There is no particular trend among undertakings regarding the calculation of the

probability distribution forecast, with some undertakings indicating that their internal
models predicted the full distribution forecast and others that only key points were
used to fit the distribution forecast. One supervisor reported that in many cases it has
been seen that the full distribution forecast came after considerable enrichment of a
distribution based on a small number of data points.

The most common method for producing the probability distribution forecast
mentioned by undertakings was Monte Carlo simulation.

Reports of the number of simulations used varied widely, from 10,000 to as many as
100,000 (the median was 25,000 simulations).

Page 109 of 153

© EIOPA 2011
For groups the responses were more homogeneous, with almost all of them stating
that the internal model outcome would be the full probability distribution.

6.2.4. Future management actions

In most cases undertakings reported taking into account the following future
management actions in their internal models:
changes in asset allocation;
changes in future bonus rates;
changes in product charges or expense charges;
changes in their reinsurance programme;
dynamic hedging; and
run-off decisions.

Some undertakings stated that in extreme scenarios, management actions may also
include exceptional actions, such as closure to new business.

One participant which was part of a group mentioned that reinsurance and run-off
actions would differ from the group internal model at local level. For another
participant management actions were not allowed at solo level.

6.2.5. Calibration

Most undertakings use the same risk measure, confidence level and time horizon for
economic capital in their internal models as defined in the Solvency II Directive:
99.5% VaR over one year. In some cases undertakings use a combination of risk
measures, which means that in addition to the VaR risk measure they use for example
a Tail Value at Risk (TVaR) risk measure as well.

Other risk measures were described as being used:

for risks that tend to occur very infrequently but are associated with large
losses;
because they are more risk-sensitive in the tails of the probability distribution
forecast;
for internal purposes; or
because the parent undertaking is under a different regulatory regime which
requires a different risk measure.

Also:
some undertakings use a higher confidence level for rating purposes;
several different confidence levels are considered for internal management
purposes;
for internal steering purposes a lower confidence level is sometimes applied at
solo level and a higher confidence level at group level; and
in some cases a longer time horizon is used because it is deemed much more
useful for running the business.

Most undertakings used the breakeven point or expected value as the attachment
point for their internal models. Other possible attachment points mentioned were the
present value and a rating agency cushion above the expected value. In terms of the
group responses, most used expected value as the attachment point.

Page 110 of 153

© EIOPA 2011
For most undertakings the same risk measures, confidence levels and time horizons
were used for all risk modules covered by the internal model.

In the opinion of one supervisor the one year 99.5% VaR of basic own funds had
become standard among undertakings. Nevertheless, it is worth mentioning the
comment of the same supervisor that the main challenge for supervisors could be
cases where the SCR calculation is based on the MCEV methodology.

The two following sections on validation tools and documentation are there to support
the figures provided on internal models by highlighting the stage of development at
which they are.

6.2.6. Validation tools

”Validation tool” means any approach designed to gain comfort that the internal
model is appropriate and reliable. Comments received from member states regarding
validation tools presented a wide range of views and in some cases they included
unique answers from some undertakings. Some common examples of validation tools
mentioned by undertakings include:
back testing;
sensitivity testing;
stress and scenario testing;
profit and loss attribution;
benchmarking; and
analysis of change.
Two groups also mentioned reverse stress tests.

It is worth mentioning some of the one-off answers regarding validation tools, for
example:
the use of consultants;
interviews with the people responsible for developing and running the internal
models as well as with the users of outputs of those internal models, including
senior management; and
some validation tools are embedded in software used by undertakings.

Regarding the stress tests undertakings made comments which in some cases were
unique to themselves. In general the design and calibration of the stress tests are
wide ranging. For undertakings which are part of groups design and calibration of the
stress tests is mostly performed at group level. In the opinion of one supervisor the
idea of reverse stress testing is not well understood within the industry and more
guidance is necessary.

On the other hand there were also comments from undertakings that validation tools
have not been chosen yet because the validation policy or definition of validation tools
was still under development.

6.2.7. Documentation

In order to analyse undertakings‟ current work on developing internal model

documentation, a representative list of key documents (which would be required for
internal model approval) was presented in the qualitative questionnaire and
undertakings could choose one of three options to describe their current status:
documentation complete - substantially fulfils the requirements;
Page 111 of 153
© EIOPA 2011
 documentation partly complete or partially fulfils the requirements; or
documentation does not exist.

In addition the documentation described both the general characteristics of the group
internal model and the differences when it is used for the solo SCR calculation.

Complete

Undertakings reported that the following documentation was mainly complete:

description of the Information Technology platform(s) used in the internal
model;
detailed description of the internal model methodology (complete at local level
as well); and
description of the contingency plans relating to the technology platform(s)
used.

The documentation least often described as complete was the validation of expert
judgment and the report of the validation test.

Undertakings using the group internal model for their solo SCR calculation reported
that the following documentation was completed:
qualitative and quantitative indicators for the coverage of risk; and
risk mitigation techniques used in the internal model.

There was only one answer that documentation concerning the evidence of the use
test at solo level was completed.

In progress

It was observed that the majority of undertakings were currently working on the
documentation and that it was partly complete. Undertakings are currently working
mainly on the:
description of the underlying assumptions;
model description and overview; or
policies, controls and procedures for the management of the internal model.

The same tendency was observed for the undertakings which used the group internal
model for their solo SCR calculation. Those undertakings also reported that their
internal model user guide was not yet complete.

Not yet commenced

Most undertakings answered that the following documentation did not exist at all at
this stage:
results of the profit and loss attribution;
model change policy and record of the major and minor changes; and
description (report) of the results of the validation tests.

For undertakings using the group internal model for their solo SCR calculation it is
worth mentioning that the following documentation was reported as not existing:
evidence of the use test;
description of the contingency plans relating to the technology platform(s)
used; and
validation of expert judgment.
Page 112 of 153
© EIOPA 2011
In summary, internal model documentation is not being prepared simultaneously with
the internal model development. The majority of undertakings mentioned that they
were still in the process of developing certain aspects of the internal model
documentation. It was found that most of the documentation concerning for example
description of the internal model, the methodology used in the internal model or
assumptions, was completed or partly completed. On the other hand most of the
undertakings do not currently possess the documentation for the validation process or
for the model change policy.

In the opinion of one supervisor a critical element concerning documentation is the

usage of external models, particularly natural catastrophe external models which
require the knowledge and skills of very specific experts.

6.3. General comparison of the internal model results with the

standard formula

In QIS5 internal model results were requested in two forms:

Method 1 - attribution of risks from undertakings‟ own internal model structure
into the standard formula structure. This part of the spreadsheet was only for
QIS5 purposes, to compare the results of the standard formula with those
derived from internal models, and did not mean that undertakings‟ internal
models had to follow the structure of the standard formula. Undertakings were
asked to provide an approximation of what the results would look like if the
internal model followed the standard formula structure.
Method 2 - to present undertakings‟ own structure of internal models.

Undertakings were strongly encouraged to supply the requested data in both forms. It
was also requested that the internal model results be aligned with (recalibrated to)
Solvency II standards (99.5% VaR over one year).

It was recognised that because internal models might be very different from each
other the quantitative information request on internal models might not be easily
completed by all participants with internal models for assessing their capital needs.

It was also noted that in the opinion of one supervisor, internal model users should
not be required to report their SCR calculation following the standard formula modular
structure, since in most cases they may use a different structure (and for partial
internal models they should only report following the standard formula for those
modules in which the standard formula is used).

Overall SCR results

The overall SCR results are regarded in this report as the most comparable figure for
analysis because it should include all risk factors and adjustments.

In QIS5 234 undertakings (about 10% of all participating undertakings) provided

overall SCR results calculated by internal models. It should be emphasised that this
meant EIOPA could not prepare detailed analysis of internal model results across the
EEA.

Page 113 of 153

© EIOPA 2011
Comparison of the internal model SCR and the standard formula SCR (based on the
small sample) demonstrates the impact which internal models have on the final capital
requirement.

Table 19: Ratio of internal model SCR to standard formula SCR

10th 25th 50th 75th 90th Weighted Standard Sample
percentile average deviation size
SCR internal model to SCR
66% 83% 91% 107% 121% 99% 0.38 236
standard formula

The table shows that the median of the SCR ratio across all undertakings was 91%
and the weighted average was 99%.

Overall, for thirteen of the nineteen countries that provided internal model results, the
median of the ratio was below 100%, with the other six countries displaying a median
above 100%.

For large and medium undertakings the median of the SCR ratio was 93%, and for
small undertakings the median was 101%.

On average in the figures provided by undertakings the internal model results were
very close to those derived by the standard formula; however, there was variation at
individual level.

In the QIS5 spreadsheet undertakings were asked to provide their economic capital
figures. The term economic capital (EC) in QIS5 referred to the capital requirement
used for the original calibration of the internal model, i.e. using the risk measure,
confidence level, time horizon and the outcome as used by the undertakings.

Table 20: Ratio of economic capital (EC) to internal model SCR

10th 25th 50th 75th 90th Weighted Standard Sample
percentile average deviation size
Economic capital to SCR
100% 100% 110% 141% 148% 123% 0.37 111
internal model

As shown in the table the median of the EC ratio across all undertakings is 110%, with
a weighted average of 123%. This ratio did not vary significantly between different
sizes of undertakings, as the EC ratio was close to 100% for small, medium and large
undertakings equally.

Groups which used full or partial internal models were also asked to provide the
capital requirements coming from these internal models.

Table 21: Ratio of group internal model SCR to group standard formula SCR
10th 25th 50th 75th 90th Weighted Standard Sample
percentile average deviation size
Group SCR internal model to
50% 60% 80% 90% 100% 80% 0.3 29
group SCR standard formula

As shown in the table the median of the group SCR calculated via internal model is
about 80% of the one deriving from the standard formula. It should be emphasised
that the number of the group submissions including solvency assessment via group
internal model is too low to allow differentiating between small, medium and large
groups.
Page 114 of 153
© EIOPA 2011
Three groups had higher SCR with their internal model than with the standard formula
(one small, one medium and one large group). Three other groups had similar results
for both methods (again one small, one medium and one large group). For the
remainder (mostly composed of small groups), the SCR calculated with the internal
model was between 46% and 90% of the SCR calculated with the standard formula.

Analysis of groups whose internal model SCR was higher than their standard formula
SCR suggests that these groups are not well diversified and are characterised by a
risk profile dominated by one type of risk. One of the groups indicated that for them
this dominant risk was not quite covered by the standard formula. In other cases, the
dominant risk was non-life underwriting risk, especially premium and reserve risk.

For the groups which had the lowest ratios of internal model SCR to standard formula
SCR, it was found that in the standard formula calculation one type of risk always
dominated. In two cases, currency risk was not identified as a risk. Furthermore, one
group which operated in many countries claimed that its assets and liabilities were
matched in terms of currency. The other risk which was not identified by those types
of groups was the counterparty default risk for reinsurance.

Allocation of the group SCR to solo entities was in most cases done on a proportional
basis calculating the contribution to the diversification effect.

6.4. Partial internal models

There are various ways to build partial internal models and the QIS5 spreadsheet
could not accommodate all of them, but nevertheless the part of their SCR calculation
covered by the standard formula should have been entirely covered in undertakings‟
submissions.

In QIS5 99 undertakings (about 42% of all undertakings which provided internal

model results) provided SCR results calculated by the partial internal models.

Table 22: Ratio of partial internal model SCR to full standard formula SCR

10th 25th 50th 75th 90th Weighted Standard Sample

percentile average deviation size
SCR partial internal
51% 80% 86% 99% 110% 82% 0.37 100
model to overall SCR

As shown in the table the median of the partial internal model SCR ratio was 86%
across all undertakings with a weighted average of 82%. One possible reason for the
90th percentile result being higher than 100% might be that in some cases the
adjustments for the loss absorbing capacity of technical provisions and deferred taxes
are not modelled in the partial internal model, despite being considered in the
standard formula calculations.

Most undertakings which plan to use partial internal models indicated that they would
use the standard formula for operational risk. Some of them will also use the standard
formula for counterparty default risk. Operational risk will simply be added to the
other risks. One interesting point is that those undertakings which plan to calculate
only operational risk with standard formula treat their models as full internal models,
when in fact they should be considered partial internal models. EIOPA‟s view is that
any model that does not cover all material risks is a partial internal model.
Page 115 of 153
© EIOPA 2011
It was mentioned by several groups that they were also intending to use the standard
formula for operational risk due to a lack of data and in the awareness that this
standard formula module lacks risk-sensitivity. For entities excluded from the scope of
the internal model (for example when they are immaterial from the group‟s point of
view), groups mentioned that they intended to use the standard formula.

The most common risk modules that undertakings plan to model are non-life
underwriting risk, market risk and life underwriting risk. In some countries
undertakings emphasised that they planned to develop an appropriate methodological
framework for the other risks in the later stages of building their internal models. As
far as the sub-modules were concerned, it seemed that undertakings predominantly
intended to replace the natural catastrophe risk and premium and reserve risk in the
non-life underwriting risk module with their partial internal model.

One country stated that for one undertaking the partial internal model will be used
wherever the risk is significant, based on the materiality criterion set by that
undertaking as ±5% of the SCR at the valuation date.

Undertakings which intend to use partial internal models under Solvency II mainly
reported that they plan to use the standard formula correlation matrix, some stating
they would replace the standard formula parameters with their own. EIOPA‟s view is
that this latter approach would not be allowed.

Undertakings also reported the following methodologies for integrating partial internal
models with the standard formula:
a variance-covariance matrix, for example for the risks not covered by the
standard formula;
using one large correlation matrix to aggregate all risks;
some sub-modules will be stochastically modelled all together; and
in only two countries, undertakings mentioned that they would be using copulas
to aggregate partial internal models with the standard formula.

One supervisor suspects that more elaborate aggregation techniques will evolve as
undertakings go through the process and further guidance is given.

Undertakings did not generally provide details about how they specified major
business units. One country commented that „major business unit‟ was defined by one
undertaking through consultation with the relevant members of management. One
supervisor expected that undertakings‟ approach to partial internal models will adjust
in response to the final outcomes on the standard calibrations (especially with regard
to the catastrophe risk module).

Page 116 of 153

© EIOPA 2011
7. MCR

7.1. MCR calculation

In general no major practical difficulties in calculating the MCR were mentioned by
undertakings. There were a number of comments about the lack of risk sensitivity in
the MCR, with a small minority of countries proposing it be set to a fixed proportion of
the SCR. One country commented that there should be a separate MCR calculation for
health.

7.2. MCR corridor

The MCR was subject to a corridor of between 25% and 45% of SCR, in order to
ensure an adequate ladder of supervisory intervention in all cases. In the absence of
this corridor the distribution of MCRs calculated would have been as follows:

Graph 50: Linear MCR as a proportion of SCR

300%
250%
% of SCR

200%
150%
100%
50%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Participants

At EEA level 35% of undertakings‟ MCRs are already within the corridor and are thus
unaffected by it. The MCRs of 41% of undertakings are subject to the lower limit while
23% of undertakings‟ MCRs are caught by the upper limit, giving the combined
distribution below.

Graph 51: Distribution of the linear and combined MCR as a proportion of SCR
1200

1000
Participants

800

600

400

200

0
%

%
p
0%

%
or

%
Ca
0

Flo

<5
10
85

75

65

55

45

40

35

30

25

20

15
0

10
>1

to
to

to

to

to

to

to

to

to

to

to

to
to

5%
%

%
%

75

65

55

45

40

35

30

25

20

15

10
85

Linear result Combined result % of SCR

Page 117 of 153

© EIOPA 2011
7.3. AMCR

As well as the adjustment for the corridor, there was also a requirement that the MCR
be greater than the AMCR (Absolute Minimum Capital Requirement) levels set for non-
life, life and composite undertakings respectively. For almost 15% of participants
across Europe, this resulted in a final MCR above the 45% cap, and for 6.6% of
undertakings, this final MCR was higher than their SCR.

Graph 52: Combined MCR as a proportion of SCR

100%

80%
% of SCR

60%

40%

20%

0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Participants

Graph 53: The corridor and AMCR effect on MCR

15%

30%

20%

35%

Above corridor At cap level In corridor At floor level

Some countries were concerned by this last group, and questioned how the ladder of
supervisory intervention would apply if there was only the hard limit of the MCR,
without the additional limit of the SCR above it.

Page 118 of 153

© EIOPA 2011
Some supervisors commented that a number of the undertakings which failed to meet
their MCR were undertakings which would have been able to meet their calculated
MCR, were it not for the AMCR coming into play and raising this higher.

Also on the AMCR, a few countries proposed that clarification on indexation and
exchange rates for non-Euro AMCR will be needed going forwards.

7.3.1. Composite AMCR

There were some comments that the AMCR floor for composites (the sum of the life
and non-life AMCR) was not consistent, and should be revised, especially in light of a
number of undertakings being clarified as composites under the new regime. It was
felt to be particularly penal for undertakings which only had a very small amount of
non-life business.

Page 119 of 153

© EIOPA 2011
8. Own Funds

8.1. Composition of Own Funds

Under Solvency II there are two types of own funds: basic own funds on the balance
sheet and ancillary own funds which are items which can be called up to absorb
losses. Basic own funds comprise the excess of assets over liabilities which is broadly
represented by ordinary share capital, the equivalent for mutual and mutual type
undertakings and reserves together with subordinated liabilities.

The tiering system categorises the own funds items according to their loss absorbing
characteristics. Various adjustments are made to take into account restrictions on the
availability of own funds items, in order to arrive at the amount available to meet the
Solvency II capital requirements. A system of limits establishes the amount of own
funds eligible to meet the SCR and MCR respectively.

Available own funds amount to €921bn in total. Of this €846bn represents the highest
quality Tier 1, which does not have any restrictions on its use to meet the SCR/MCR. A
breakdown of the available own funds is below. From this it can be seen that the
majority is basic own funds with ancillary own funds representing only 1.3%.

Graph 54: Composition of available own

100%
90%
funds
80%
70%
60%
50%
Solo
40%
Group
30%
20%
10%
0%
Tier 1 Tier 1 Tier 2 Basic Tier 2 Tier 3 Basic Tier 3
Unrestricted Restricted Own Funds Ancillary Own Funds Ancillary
Own Funds Own Funds

8.1.1. Composition for groups

When applying the accounting consolidation-based method, group own funds are
determined by using the group balance sheet valued according to the QIS5
specifications. It therefore includes all insurance and reinsurance undertakings, SPVs,
insurance holding companies and ancillary entities, both EEA and non-EEA. Own funds
related to other financial sectors amount to less than 3% of total group own funds.
This small share is partly due to the fact that some financial conglomerates did not
include their banking sector entities in the QIS5 exercise. Group own funds derived
from the insurance sector were adjusted to take account of the non-availability of
Page 120 of 153
© EIOPA 2011
certain items and determine the level of group available own funds (see groups
section of this report for further details).

A breakdown of group own funds is in table xOF1 in the annex. Ancillary own funds
represent less than 1% for the groups sample. The most significant difference as
compared with the data for solo undertakings is the relative proportion across the
tiers.

8.2. Unrestricted Tier 1

This represents the highest quality own funds and accounts for 91.9% of available
own funds (for groups 81.5%).

QIS5 analyses the components of unrestricted Tier 1 by extracting the accounting

balance sheet values for ordinary share capital, mutuals‟ initial fund, share premium,
retained earnings and other reserves appearing in the accounts.

Ordinary share capital (net of own

Graph 55: Split of BOF items (solo) shares)
The initial fund (less item of the same
type held)
Share premium account

Retained earnings including profits from

the year net of foreseable dividends
Other reserves from accounting balance
sheet
Reconciliation reserve
EEA

Surplus funds

Expected profit in future premiums

Preference shares

Subordinated liabilities

Subordinated mutual member accounts

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Other items not specified above

[G]:lMinority interests

Page 121 of 153

© EIOPA 2011
 Graph 55: Split of BOF items (groups) Ordinary share capital (net of own
shares)
The initial fund (less item of the same
type held)
Share premium account

Retained earnings including profits from

the year net of foreseable dividends
Other reserves from accounting balance
sheet
Reconciliation reserve

GP
Surplus funds

Expected profit in future premiums

Preference shares

Subordinated liabilities

Subordinated mutual member accounts

Other items not specified above

-20% -10% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
[G]:lMinority interests

The above graphs set out the contribution of the different components of unrestricted
Tier 1 highlighting the relationship between share capital and share premium and the
reserves held by undertakings.

Surplus funds (as described in Article 96 of the Framework Directive) form a separate
item in unrestricted tier 1 amounting to €69bn for all solo undertakings, which is
significant for undertakings in a number of countries.

The reconciliation reserve is introduced in Solvency II to ensure that the basic own
funds can be reconciled back to the excess of assets over liabilities. It demonstrates
the effect of moving from the accounting balance sheet to the QIS5 balance sheet
(although due to tiering effects it will not always be exactly comparable). It is not split
up for classification purposes and counts as unrestricted tier 1.

Under QIS5 separate disclosure was made of an item representing expected profits
included in future premiums (EPIFP) – discussed further below. This would usually
form part of the reconciliation reserve, and so as a result of this separate disclosure a
negative adjustment is necessary in order to balance the reconciliation reserve.

So both the net effect of balance sheet movements and the adjustment for EPIFP
contribute to the make-up of the reconciliation reserve and thus to Tier 1 own funds.
At EEA solo level the positive value of €110bn is driven by a €241bn reduction in
technical provisions, offset against a decrease in asset values and an increase in other
liabilities and reduced by the EPIFP adjustment. A different pattern emerges in the
groups sample where the reconciliation reserve is negative (-€24bn). The net effect of
balance sheet movements is less than the EPIFP adjustment. This is because the
reduction in technical provisions of €102bn is offset by a comparatively larger
decrease in asset values for the groups sample.

Page 122 of 153

© EIOPA 2011
8.3. Comparison with Solvency I
Having examined the difference between the accounting balance sheet and QIS5 as
depicted by the reconciliation reserve, it is also useful to compare own funds under
QIS5 with Solvency I. Total basic own funds before adjustment under QIS5 were 27%
higher than the Solvency I total. Informed by the presence of the reconciliation
reserve it can be seen that changes in the valuation basis, particularly for technical
provisions, are behind this increase. This is in conjunction with smaller changes
affecting share premium and subordinated liabilities, and changes in surplus funds and
retained earnings which offset each other in magnitude.

Out of 167 groups which participated in QIS5, 146 provided information on own funds
eligible under Solvency I. Making the same comparison as above on the groups data,
the increase in basic own funds under QIS5 against the Solvency I position is 38%.
Again there are various items where increases and decreases largely offset one
another, leaving the increase largely driven by the decrease in technical provisions.

8.4. Other paid-in capital instruments

Under QIS5 the classification of other paid-in capital instruments (preference shares,
subordinated mutual members‟ accounts and subordinated liabilities) depends on the
presence of certain features to enable the instrument to meet the relevant criteria for
each tier. In the future these items will need to exhibit stronger loss absorbency
characteristics in order to qualify as own funds and those eligible for Tier 1 (restricted
to 20% of Tier 1) will need to demonstrate effective going concern loss absorbency.

8.4.1. Transitional provisions for own funds

Transitional provisions are key to the implementation of any new regime for own
funds. While there will always be a policy objective of achieving full compliance with a
new regime on an expeditious basis, an equally important objective is the prevention
of market dislocation and the promotion of a smooth transition that has regard to the
tenor of current issuance. Discussions around transitional provisions for Solvency II
own funds are still underway with the aim of achieving this balance and so the
transitional provisions tested in QIS5 are not final.

The contribution of hybrid capital and subordinated debt instruments to the capital
base of the European insurance sector is a material amount in absolute terms, as well
as being significant in terms of the debt capital markets more generally. However the
impact is not uniformly distributed across the EEA, with a number of countries where
undertakings make little or no use of this form of own funds.

QIS5 was intended to test the application of the Solvency II criteria for basic own
funds on the basis that all current capital instruments were measured for compliance
as though Solvency II were in force. This was referred to as “without transitional
measures”. An additional scenario tested the position against the proposed criteria for
transitional measures. The difference between the two was intended to demonstrate
the extent to which transitional provisions are required in respect of capital
instruments.

Page 123 of 153

© EIOPA 2011
The quantitative results from QIS5 in this area do not give a clear picture.
Undertakings across the EEA have reported existing hybrid and subordinated debt
instruments in Tiers 1, 2 and 3 in the “without transitional measures” scenario.
Supervisors in countries where these instruments are currently in issuance are of the
view that these items should not qualify as basic own funds under this scenario. The
technical specifications appear to have been interpreted very widely and somewhat
optimistically by undertakings in arriving at the opposite view, notwithstanding their
identification of transitional measures as a significant issue for them.

The data are further confused because the amounts reported for most categories of
own funds under the “with transitional measures scenario” are lower than those
without. This improbable result arises because not all undertakings have completed
the “with transitional measures” submission. The basis of comparison between the two
sets of data is therefore undermined.

However the amount of subordinated liabilities currently reported at the solo level by
QIS5 participants gives a measure of the potential impact of, and need for,
transitional provisions. The total amount of subordinated liabilities regardless of tier is
€48bn. Preference shares used by some undertakings in certain countries amounted
to €0.9bn and subordinated mutual members accounts to €0.2bn. The total for other
paid-in capital instruments is €49bn representing 5.2% of available own funds.

The following graph sets out whether these instruments are dated or undated and
with or without issuer call features.

Graph 56: Features of other paid-in capital instruments

11.5%
Dated
instruments

40.4%
Undated with
issuer call
feature

48.1%
Other undated

Groups

Some of these capital instruments reported at solo level may be externally issued but
a significant proportion is likely to be intra-group own funds. The need for transitional
provisions must therefore also have regard to the position for groups

For the groups data the other paid-in capital instruments amounted to €83bn, of
which subordinated liabilities represent nearly €82bn and preference shares €1.5bn.

Page 124 of 153

© EIOPA 2011
Consistent with the position under QIS4, the proportion of group own funds
represented by subordinated liabilities is, at 15.6%, higher than at solo level. This is
due to the fact that groups raise capital at holding company level or via a capital-
raising subsidiary within the group and then down-stream it to other group companies
in the form of higher quality own funds. Table xOF5 in the annex provides a similar
analysis of the instruments at a group level.

8.4.2. Features of other paid-in capital instruments

QIS5 sought to establish whether any existing instruments include the types of
principal loss absorbency mechanism which are fundamental to the going concern loss
absorbency characteristic for Tier 1 under Solvency II.

Write-down mechanisms and conversion features

A number of arrangements were described involving various features but none of

these appeared to provide for the principal of the instrument to be written down.

Only one example of a conversion feature was provided. This involved the conversion,
at a given trigger, of a subordinated liability into a profit sharing certificate
representing the full principal and accrued and unpaid interest.

The feedback set out above confirms that the market would need to develop new
forms of instrument in order to secure Tier 1 treatment under Solvency II. The
responses also reinforce the expectation that if any existing instrument is to qualify
for the “other paid-in capital instruments” category in Tier 1 (the 20% bucket), it will
do so based on satisfaction of the transitional measures criteria in their final form.

ACSM

The qualitative questionnaire sought information from participants on the extent to

which existing capital instruments contain alternative coupon satisfaction mechanisms
(ACSM). For many countries there was nothing to report under this heading. Details
were reported by six countries, with four referring to one example each and two
others citing a small number of examples.

The ACSM described appear to have similar characteristics, generally providing for the
issue or sale of ordinary shares to satisfy the deferred interest. In addition some also
provide for the issue of alternative securities junior to or pari passu with the
instrument in question. Helpful details on the conditions under which the ACSM
operates were also provided in some cases.

The responses make clear that ACSM are not common throughout the EEA, but that
their characteristics are in line with the general understanding of market practice in
this area. A key aspect is that ACSM generally come into play at the point at which
coupons are resumed and not at the point of deferral. This has been criticised as it
may create a build-up of a commitment to issue ordinary shares at a time when
recapitalisation might be seen as a higher priority than compensating hybrid debt-
holders.

Whether a revised form of ACSM would be compatible with full flexibility on coupons
and coupon cancellation upon breach of the SCR will depend on the detail of the
Solvency II criteria.
Page 125 of 153
© EIOPA 2011
8.4.3. Impact of tiering and limits

Having classified own funds into the respective tiers according to their satisfaction or
not of the relevant QIS5 criteria, the limits operate in order to establish the amount of
own funds eligible to meet the SCR and MCR respectively. The spreadsheet for QIS5
permitted undertakings to approach the calculation of eligible Tier 2 and Tier 3 in two
ways. The so-called “top down” approach maximised the use of Tier 2 within the
maximum 50% of the SCR and only when Tier 2 was fully utilised within this limit was
any Tier 3 considered. In contrast the “bottom up” approach maximised the use of
Tier 3 up to the limit of 15% of the SCR and then brought in any Tier 2 for
consideration against the combined 50% limit. This approach was adopted to simplify
the operation for QIS5; in practice under Solvency II undertakings would make their
own choice as to the balance between Tier 2 and Tier 3.

The overall impact of the limits is difficult to assess because of the issues relating to
the inappropriate inclusion of instruments under the “without transitional measures”
scenario and the more limited completion of the “with transitional measures” scenario.
However the following broad conclusions can be drawn:

From an overall perspective and on the basis of the data provided the choice of
approach for applying the limits to tier 2 and tier 3 was not significant.
Setting aside the issues regarding the inclusion of restricted tier 1, the 20%
limit appears to cause significant restriction under the without transitional
measures scenario.
There are clearly anomalies in the with transitional measures data because the
amount for unrestricted tier 1 varies even though it is not subject to limits and
the choice of approach for tiers 2 and 3 appears to impact the eligibility of
restricted tier 1.
On the evidence of the data reported tier 2 limits do not appear to have a major
impact.
Capital instruments were also included in Tier 3 in both scenarios. Again the
inclusion of these items under the without transitional measures scenario is
unlikely and the technical specifications did not provide for transitional
measures under Tier 3.
The 15% Tier 3 limit appears to have a greater impact.

In general the pattern for the groups sample follows that of the solo data - particularly
in that there is little if any difference between the top-down and bottom-up
approaches. Restrictions due to limits can be observed in each of the three tiers. The
greater significance of other paid-in capital instruments is marked, even though their
allocation to tiers is unlikely to be reliable.

8.5. Adjustments to Basic Own Funds

These can arise in relation to ring-fenced funds, restricted reserves, participations in

financial and credit institutions and net deferred tax assets.

The nature and quantum of the various adjustments are set out in the chart below.

Page 126 of 153

© EIOPA 2011
 Graph 57: Adjustments to Basic Own Funds
960
100%
99.99%
940
97.99% 97.22% 97.22% 97.22%
920
97.00%
900
Tier 3
€bn

880
Tier 2
860
Tier 1
840

820

800
Basic Own Restricted Participations Ring fenced Restricted Net deferred Basic Own
Funds before reserves in credit and funds (excess reserves tax assets - Funds after
adjustments deduction financial own funds relegation relegate from other
institutions over notional Tier 1 to Tier 3 adjustments
SCR)

8.5.1. Ring-fenced funds

The QIS5 Technical Specifications contained information designed to help

undertakings identify when they have ring-fenced funds – that is, where own funds
items have a reduced capacity to fully absorb losses on a going concern basis owing to
their lack of transferability within the undertaking because the restricted own funds
can only be used to cover losses:
on a defined portion of the undertaking‟s (re)insurance contracts; or
in respect of particular policyholders or beneficiaries; or
in relation to particular risks.

These restrictions may arise from statutory constraints or aspects of product design
applicable to many undertakings, or in some jurisdictions from specific contractual
terms relating to one or a small number of situations.

In order to gain insight into the types of ring-fenced arrangements, undertakings were
asked to describe the arrangements giving rise to ring-fenced funds and the nature of
the restrictions which apply.

A majority of countries reported that there were no ring-fenced arrangements within

their territory. Six countries reported one or very few ring-fenced arrangements, and
five reported several or many ring-fenced funds. Three countries had reservations
about data quality and/or wanted greater clarity on what constitutes a ring-fenced
fund to enable them to be confident about industry feedback.

Countries reporting ring-fenced funds did so on the basis that transfers out of the fund
were restricted (or not allowed at all). Several countries reported with-profit funds.
These have restrictions on transfers from the with-profit funds to shareholders.
Individual pension plans were considered as ring-fenced funds in a number of
jurisdictions.

In some cases where some undertakings or part of the industry concluded that there
was no ring-fencing, this did not align with the views of the local supervisory authority
or other parts of the industry. Conversely, there was one case where a ring-fenced

Page 127 of 153

© EIOPA 2011
arrangement reported by the industry was deemed by the supervisory authority not to
be ring-fenced.

The above results suggest that participants and supervisors would benefit from
greater clarity about the characteristics of ring-fenced funds. This should lead to
greater consistency in application. It should be emphasised that specific arrangements
need to be individually assessed on the basis of the detailed contractual or legislative
provisions that apply, in order to decide whether or not they are ring-fenced.

In terms of the calculations carried out by those undertakings identifying ring-fenced

funds, it is clear from a number of supervisory comments that there are concerns as
to manner in which the calculations were carried out or carried through to effect
adjustments in the SCR and/or the own funds. Undertakings appear to have had
difficulty recording the calculations consistently through the relevant parts of the
spreadsheet.

QIS5 data reveal that 80 undertakings identified 218 ring-fenced funds spread across
eight countries amounting to €17.3bn of assets. Once a ring-fenced fund has been
identified it does not necessarily mean that there will be an adjustment restricting the
own funds within the ring-fenced fund. Only if there are own funds in excess of the
notional SCR will an adjustment be necessary.

Five countries account for €7.03bn in terms of adjustment to own funds. One further
country reported a positive adjustment reducing the overall impact to €6.96bn. As
there should never be a positive adjustment this further emphasises the need to
promote a better understanding among undertakings.

8.5.2. Restricted reserves

The QIS5 Technical Specifications defined restricted reserves as reserves which might
be required, whether under national law or under the specific statutes/articles of an
undertaking, to be established and used only for certain prescribed purposes. The
technical specifications expressly stated that equalisation provisions would form part
of the reconciliation reserve and should not be treated as restricted reserves.

Because of these requirements, the amount of any own funds in excess of amounts
being used to cover related risks within a restricted reserve is not available to absorb
losses elsewhere in the undertaking on a going concern basis. Therefore undertakings
were required to deduct any excess from Tier 1 own funds, and only to recognise it as
Tier 2 if it was available to meet all losses in a winding up.

Undertakings were in some cases confused about identifying restricted reserves;

many of the reported items were obviously erroneous and had been highlighted as
such in supervisory comments. Examples of this included non-capital stock deposits,
equalisation provisions (identified by undertakings in three different countries) and
funds for the purchase of own shares.

Other items reported as restricted reserves appeared more like ring-fenced funds in
nature, including with-profit funds and reserve for guarantees funds. Also,
undertakings in two countries reported collateralised assets as restricted reserves.
These should not be adjusted for so far as they match liabilities, but any over-
collateralisation would more appropriately be treated as ring-fenced funds.

Page 128 of 153

© EIOPA 2011
The only significant restricted reserves on which supervisory authorities commented
were reported by 31 non-life undertakings in one jurisdiction. Another country
referred to risk equalisation funds held by life undertakings and natural perils funds
held by non-life undertakings as the key restricted reserves in their market.
Additionally several countries had undertakings which reported reserves whose names
suggested they could potentially be restricted reserves, or alternatively might be
provisions or liabilities. These included four countries with undertakings reporting legal
reserves/reserve funds, two countries with undertakings reporting risk equalisation
funds/reserves, and undertakings within individual countries also reporting
contingency funds, voluntary reserves, statutory reserve funds and legal development
funds. In none of these cases did the supervisory comment indicate that these were
restricted reserves; in one case they commented that a legal reserve was not a
restricted reserve.

The qualitative comments appear to be borne out by the data.

An amount of €0.1bn was deducted from Tier 1 from undertakings in twelve countries.
Under the terms of the technical specifications this indicates that these reserves
should not be regarded as meeting the criteria for loss absorbency in a going concern
(Tier 1) or in a winding-up (Tier 2).

More significant was the amount of restricted reserves which were relegated from Tier
1 to Tier 2. Under this treatment the amount of the restricted reserves in excess of
the related risks is considered available to absorb losses more generally in a winding-
up. In this category €5.71bn was reported across eight countries of which €5.50bn
was accounted for by the single jurisdiction described above.

8.5.3. Participations in financial and credit institutions

The amount of the deduction was €18.6bn reported by 95 undertakings in 20

countries.

8.5.4. Net deferred tax assets

This adjustment is of a different nature to those discussed above, as it is made to

account for the lack of immediate availability of the own funds, rather than restrictions
on their usage as in the previous cases. The relegation of net DTA from Tier 1 to Tier
3 was reported by 408 undertakings distributed across all countries. The amount of
€9.56bn represented 56% of available basic Tier 3 of €17.02bn. As noted previously
the 15% limit for tier 3 does come into play. Assuming no other tier 3 items it is likely
that for some undertakings net DTA would be subject to restriction.

8.6. Ancillary Own Funds

Ancillary own funds (AOF) as off-balance-sheet capital represent one of the key
developments of Solvency II. Under the Solvency I regime certain items form part of
the available solvency margin but will be treated as AOF under Solvency II. These are
generally specific to certain market sectors – a good example is the calls which
mutuals may make on their members. A more general framework for off-balance-
sheet items extends the potential for undertakings to use such items although it is not
clear to what extent they will seek to make use of AOF once Solvency II is
implemented. The nature of AOF means that there must be the safeguard of
Page 129 of 153
© EIOPA 2011
supervisory approval, as specified in the Framework Directive, before such items can
be counted towards the SCR.

AOF are classified as ancillary tier 2 if on being called up and paid in they become tier
1 or otherwise as ancillary tier 3.

Ancillary tier 2 amounted to €11.6bn spread over thirteen countries of which €10.5bn
was accounted for by three countries. Ancillary tier 3 was much less significant, with
€0.1bn reported over four countries.

The graph below provides a breakdown of these amounts by type of ancillary own
funds. A significant amount of the ancillary tier 2 represents letters of credit and
guarantees held in trust by an independent trustee, as envisaged under article 96 (2)
of the Framework Directive, together with supplementary calls by mutual and mutual
type organisations, also referred to in article 96.

Graph 58: Composition of ancillary own funds

120% Other items currently
eligible to meet
requirements under
100% Solvency I
Mutual calls for
supplementary
contributions (OF31(c))
80%
Mutual calls for
supplementary
60% contributions (OF31(b))

Letters of credit and

40% guarantees held in trust
(Article 96(2))

20%
unpaid share capital or
initial fund that has not
been paid up
0%
Tier 2 Tier 3

In addition to the quantitative results for AOF, QIS5 sought qualitative data on:
existing items currently recognised as own funds but which would be AOF under
Solvency II;
existing items not counting towards the solvency margin but which might be
AOF; and
new arrangements for which AOF approval might be sought.

As well as providing a perspective on the potential contribution of AOF to the Solvency

II regime these questions were also intended to assist supervisory authorities in
assessing the likely calls on their time and resources for the approval of AOF during
and after implementation.

For many countries there are no existing arrangements that might become AOF and
for these, and indeed others where existing arrangements would become AOF,
undertakings seem to envisage little concrete planning to develop new arrangements.
There were isolated examples across a few countries suggesting the development of

Page 130 of 153

© EIOPA 2011
letters of credit and the use of unpaid share capital subscribed by a parent
undertaking.

8.7. EPIFP
QIS5 required the identification and calculation of an amount representing expected
profits included in future premiums (EPIFP) and its disclosure as a separate item
under unrestricted Tier 1. For this purpose future premiums are those taken into
account as part of the cash inflows used to determine technical provisions under
Solvency II. The identification of an EPIFP amount as a component of the excess of
assets over liabilities within Tier 1 is intended to inform the continuing policy debate
and Level 2 negotiations.

Industry does not generally have the data with which to calculate EPIFP as it is now
understood. In order to provide a quantification as part of QIS5 a proxy methodology
was developed in liaison with industry bodies which utilises the lapse risk methodology
already specified for the SCR but re-calculates this for EPIFP purposes on the basis of
a 100% lapse. Application of the proxy required undertakings to hold all other
assumptions unchanged even if this involved creating artificial calculations of a paid-
up amount for policies for which no paid-up amount arises or which would be void or
cancelled if premiums were not paid in practice.

EPIFP attracted a significant level of comment during QIS5. The comments fell into
the following categories:
In a number of countries a proportion of life undertakings did confirm they were
satisfied with the clarity of the methodology.
But lack of clarity and difficulty in arriving at paid-up assumptions – particularly
on the part of non-life undertakings – were generally cited across Member
States.
Calculations were described as time-consuming and burdensome and of
questionable benefit.
Undertakings questioned the concept and this affected the manner in which
they engaged with the calculations or whether they attempted them at all.
Supervisory authorities generally confirmed these comments and some
suggested that the calculation should not be performed. A common theme was
the potential significance of the amounts.

Within the spectrum of these reactions it is clear that a significant number of

undertakings did not complete the calculation. Some set the result to nil and at the
other extreme some undertakings set the amount at or greater than the amount of
technical provisions and/or basic own funds.

While the results of the exercise can be used to inform the policy debate in this area
the nature and extent of the commentary means that care should be exercised in the
use of these data. In particular it is unlikely that the data obtained can be
extrapolated safely.

Of 2520 undertakings participating in QIS5 745 (29%) identified EPIFP. Across the
different countries there was significant variation in levels of completion, ranging from
nil to 80% with an average across Member States of 34%. Five countries accounted
for 70% of the EPIFP total amount of €83.7bn. Putting these amounts in context the
weighted average percentage of EPIFP to Tier 1 among undertakings was 20% with a
median of 14%. However it should be noted that the Tier 1 attributable to the

Page 131 of 153

© EIOPA 2011
undertakings which did identify EPIFP represents 45% of Tier 1 for QIS5 as a whole
indicating that this group is drawn from larger undertakings.

Graph 59:

Amount of EPIFP divided by Tier 1

90%
80%
70%
60%
50%
40%
30%
20%
10%
0%

AT

EEA
PL

NL
LV
LU
LT

MT

EE

CY

BE

DE

SE
HU

PT
BG
CZ

FR
GR

ES
SK
UK

DK

GP
NO

RO
IE

IT
LI

SI

FI

IS
Not surprisingly EPIFP is a more important component of own funds for life and health
insurers as compared to non-life insurers. As emphasised above though, results
should be treated carefully, as they differ greatly between undertakings and countries.
Furthermore, from the qualitative responses it appears that many non-life
undertakings did not engage with this part of QIS5 or assumed it was relatively
insignificant.

Graph 60: Ratio of EPIFP to total tier 1 by category of

undertaking
35%
30%
25%
20%
15%
10% Solo
5% Groups
0%
l

e
ive

th

e
Al

Lif
sit

Lif

nc
al
pt

po

n-

ra
He
Ca

su
No
m

in
Co

Re

Further context can be provided by considering the data for EPIFP provided by the
groups taking part in QIS5.

Out of 167 respondents, 96 submitted data related to EPIFP (57% of the total). The
total amount reported was €61.5bn representing on average 16% of Tier 1. However,
the 96 groups which did report EPIFP represent 76% of the groups‟ total eligible tier
1, reinforcing the finding at solo level that the data reported are derived from the
larger undertakings and larger groups. Nevertheless, the reduced sample size means
that many of the caveats set out above should also apply to the groups data. A
number of groups stated that they found it difficult to carry out the calculation
envisaged in the QIS5 Technical Specifications.
Page 132 of 153
© EIOPA 2011
9. Groups

9.1. Participation and methods tested

167 groups from eighteen countries participated in the groups part of the QIS5
exercise at centralised group level. Compared to the participation in QIS4 (106
groups), this is a significant improvement.

Groups were asked to test the methods envisaged in the Solvency II directive for
calculating group capital requirements. In particular, they were required:
to test both the accounting consolidation-based method and the deduction and
aggregation (D&A) method (calculating it with both Solvency II and local rules
for non-EEA entities);
to provide data, if relevant, related to the internal model calculation; and
to provide data, on an optional basis, on the application of a combination of
methods.

The results are reported in the table below.

Table 23: Groups participation by methods tested

EEA EEA EEA Total
groups groups subgroup(s) of respondents
without with non- non-EEA
non-EEA EEA groups
entities entities
S1 - Current calculations 109 38 4 151
SII – Consolidated method 120 41 5 166
SII – D&A (SII applied to the 99 36 3 138
non-EEA entities)

SII – D&A (local rules applied Not 27 Not relevant 27

to the non-EEA entities) relevant
SII – Internal model 10 17 2 29
SII – Combination of methods 1 4 0 5
121 41 5 167
Sample size

As shown in the table, all groups apart from one tested the accounting consolidation-
based method. Far fewer groups tested the deduction and aggregation method, citing
time constraints, lack of familiarity with the D&A method, and the need for further
guidance in the technical specifications as reasons for this. Nevertheless the sample is
large enough for conclusions to be drawn about the differences between the two
methods, and notably almost all groups with non-EEA entities have provided
calculations for those entities according to Solvency II rules.

Finally, only five groups have tested a combination of methods, all using different
approaches to combine them.

As regards the quality of the submissions, the assessments provided by the group
supervisors varied from one group to another, depending in part on the group‟s
participation in the previous QIS exercises. In most cases, the quality was assessed as
sufficient. Areas where groups have encountered major difficulties relate to the

Page 133 of 153

© EIOPA 2011
valuation and absorbing effects of deferred taxes and future discretionary benefits at
group level, the treatment of ring fenced funds and intra-group transactions.

The table below outlines the number of participants by size: large groups were those
which had total assets greater than €90bn, and small groups as those with total
assets less than €30bn.

It is important to note the high participation rate among small groups, which explains
most of the improvement in the participation rate between QIS4 and QIS5.

Table 24: Groups participation by size

Total Large Medium Small
S1 - Current calculations 151 17 22 112
SII – Consolidated 166 17 23 126
method
SII – D&A (SII applied to 138 15 22 101
the non-EEA entities)
SII – D&A (local rules 27 13 4 10
applied to the non-EEA
entities)
SII – Internal model 29 9 6 14
SII – Combination of 5 4 0 1
methods
Sample size 167 17 23 127

9.2. Comparisons of the various methods and principal results

The following table shows that the vast majority (96.1%) of the capital requirements
reported under QIS5 come from the core insurance business. 28 groups reported
capital requirements for non-controlled participations, which on average represented
1.4% of their total group SCR. 65 groups reported figures for other financial sectors,
which on average accounted for 5.4% of the total group SCR for those groups. It
should be noted that most bancassurance groups only reported figures relating to
their insurance business. The surplus amounts indicated throughout this report refer
to the total group SCR.

Table 25: composition of the group SCR

Weighted average Group SCR SCR* Non-controlled Other financial
participations sectors
Global statistics 100% 96.1% 1.4% 5.4%
Sample size 166 166 28 65

9.2.1. Total evolution of the surpluses between Solvency I and II

Before analysing the surpluses, it is worth mentioning the following caveats. This
analysis assumes transitional measures on own funds: even when groups did not
provide figures both with and without transitional measures, it was noted that in most
cases they allocated own funds as if transitional measures were in place (by
classifying most of their own funds into tier 1). Moreover, sixteen groups which had
entities in non-EEA countries did not provide data on own funds according to local
rules, leaving only eleven groups for which the surplus under the deduction and
aggregation method with local rules could be analysed.
Page 134 of 153
© EIOPA 2011
The table below gives a comparison of the QIS5 surplus against Solvency I under the
different calculation methods.

Table 26: Ratio of surplus under QIS5 to surplus under Solvency I

Assumptions 10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
SII -
Accounting
-78% 18% 75% 138% 259% 57% 359% 146
consolidation-
based method
SII - D&A (SII
applied to the
-131% -20% 59% 120% 291% 28% 455% 124
non-EEA
entities)
SII - D&A (local
rules applied to
-38% 15% 63% 95% 113% 52% 68% 11
the non-EEA
entities)
SII - Internal 106%
-68% 14% 102% 138% 177% 753% 26
model
SII –
Combination of 48% 49% 59% 72% 118% 75% 42% 5
methods
SII – Highest
-84% 41% 85% 148% 290% 98% 493% 147
surplus

For each method, the ratio of QIS5 to Solvency I surplus varies a lot between groups,
as shown by the high standard deviation figures. Moreover, it was observed that some
ratios were negative, which can be explained by negative surpluses (deficits) under
QIS5 (as only two deficits were reported under Solvency I).

Accounting consolidation-based method

For groups using the accounting consolidation-based method with the standard
formula there is a reduction in group surplus of around €86bn compared to Solvency
I, from about €200bn to €114bn, resulting in a weighted average of QIS5 surplus to
Solvency I surplus of 57%.

For groups that submitted internal model results, there was an increase in surplus of
about €6bn moving from Solvency I to QIS5 resulting in a weighted average surplus
of 106%. However, it should be noted that there was substantial variation in the
individual groups‟ results.

Deduction & aggregation and use of local rules

On average, when groups applied the deduction & aggregation method rather than the
accounting consolidation-based method, there was a significant reduction in surplus.
This is due to the non-recognition of diversification effects under this method.
Page 135 of 153
© EIOPA 2011
For groups with entities in non-EEA countries, the application of local rules instead of
Solvency II rules had a significant impact on surplus, as shown by the change in the
extent of the negative surplus for the 10th percentile of groups.

In order to estimate the overall impact of the application of local rules for all groups
which reported capital requirements for the deduction & aggregation method using
local rules (27 groups), and not only for the eleven groups which provided own funds
calculated according to local rules, it was assumed that for the other groups their own
funds under local rules were equal to their own funds valuated with Solvency II rules.
This approximation is acceptable, since for groups which did submit both results there
were no significant changes in the level of own funds when using local or Solvency II
rules. Using this approximation, the overall impact of using local rules for non-EEA
entities under the deduction and aggregation method is about €45bn, justifying a
transitional measure with review clause for a few third countries that have not yet met
the equivalence test.

Five groups submitted results using a combination of methods, applying a mixture of

the accounting consolidation-based method, deduction and aggregation and partial
internal model to different parts of the group. For these groups the weighted average
of QIS5 surplus to Solvency I surplus was around 75%.

Conclusions

Based on the above analysis (see also the table below), the group surplus eligible own
funds under QIS5 were €86bn lower than under Solvency I if the accounting
consolidation-based method with the standard formula was used. However, the
surplus would only be €3bn lower if group internal models (partial or full) were used
at their current stage of development and either equivalence were granted for third
country jurisdictions or transitional measures were put in place allowing the use of
local rules under deduction and aggregation for third countries. The table below shows
that the impact of the different Solvency II calculation methods predominantly
affected large groups.

Table 27: Ratio of surplus under QIS5 to surplus under Solvency I when using internal models
and local rules for third countries
(€bn) Surplus Solvency I Surplus QIS5 Sample size
Results assuming internal models were approved and local rules under D&A for third
countries were used
Large 109.4 129.5 17
Medium 26.7 18.3 21
Small 64.3 49.5 109
All 200.4 197.4 147
Accounting consolidation-based method with standard formula
Large 109.4 54.6 17
Medium 26.7 15.5 21
Small 64.2 43.6 108
All 200.3 113.7 146

Page 136 of 153

© EIOPA 2011
9.3. SCR standard formula - diversification effects

9.3.1. Diversification at SCR level

Diversification

The group diversification effects are measured as the ratio between the group SCR
calculated using the accounting consolidation-based method and the sum of the solo
SCRs. In order to have reliable results, the numerator (group SCR) and denominator
(sum of solo SCRs) of the ratio have to include all of the same entities. For
consistency, the proportional share to be used when calculating the sum of the solo
SCRs is set at the same percentage as the one used to consolidate the entities of the
group when calculating the group SCR.

Table 28: Diversification effects

10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Ratio of group SCR to sum
65% 79% 90% 96% 100% 80% 18% 138
of solo SCRs
Diversification 35% 21% 10% 4% 0% 20%

The table indicates that the diversification was on average equal to 20%. However it
should be noted that this figure varied widely between groups.

The group diversification effects above use the sum of the solo SCRs including the
capital charges on intra-group transactions. As a result these effects include both
“real” diversification effects, following the application of the standard formula to a
wider range of activities, and the effects of eliminating intra-group transactions. These
two effects can be separated and analysed.

Intra-group transactions

Comparing the sum of the solo SCRs and the sum of the solo adjusted SCRs (i.e. net
of intra-group transactions) makes it possible to capture the impact of intra-group
transactions on total diversification.

Table 29: Intra-group transactions effects

10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Ratio of sum of solo
adjusted SCRs to sum of 89% 98% 100% 100% 100% 91% 11% 140
solo SCR
Intra-group effects 11% 2% 0% 0% 0% 9%

This table shows that the impact of intra-group transactions is on average 9%.

This overall effect should be regarded with caution for two reasons: firstly the intra-
group transaction effects appeared particularly high for one medium size group (since
it had a high intra-group counterparty risk charge at solo level, which was eliminated
at group level through consolidation). Secondly, it should be noted that for many
groups no intra-group transactions have been reported, giving a ratio of nil. It could
be that these groups did not provide the adjusted data because they found it difficult

Page 137 of 153

© EIOPA 2011
to do the calculation. As a result, the figures above are likely to be an
underestimation.

“Real” diversification

The ratio between the group SCR and the sum of solo adjusted SCRs provides a
measure of “real” diversification.

Table 30: Real diversification effects

10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Ratio of group SCR to sum
70% 83% 92% 98% 100% 87% 16% 138
of solo adjusted SCRs
Real diversification 30% 17% 8% 2% 0% 13%

As shown in the table, the “real” diversification was on average 13%. The
diversification effect varied according to the size of the group; it was relatively low for
small groups (less than 5%), higher for medium groups (around 16%) and relatively
great for the largest groups (around 21%). Diversification largely depended on the
diversity of the group‟s activities and locations, which was usually more material for
large groups.

However, as mentioned previously, it should be noted that the solo adjusted SCRs
were not always calculated accurately, meaning the “real” diversification effects may
be overestimated.

It should also be noted that for some groups, the “real” diversification effects appear
to be negative. This can be explained in two ways: firstly, when calculating the group
SCR based on consolidated data, groups were required to consider all entities within
the scope of the group, including for example holding companies; however,
sometimes these entities were not included when calculating the solo requirement,
leading to a negative diversification effect; secondly, in some cases the adjustment
made for the loss absorbency of deferred taxes at group level was different to the
sum of the adjustments made at solo level, which could also result in a counter-
intuitive result for diversification.

9.3.2. Diversification at the level of the different risks

Intra-group transactions

This section provides a more in-depth analysis of the impact of intra-group

transactions, analysing the adjustments for intra-group transactions relating to the
different modules and sub-modules.

In the QIS5 exercise, groups were asked to provide the SCR calculation with
adjustments for intra-group transactions under the following modules (and underlying
sub-modules): market risk, counterparty default risk, operational risk, life
underwriting risk, health underwriting risk and non-life underwriting risk, taking into
account of their materiality. Intra-group transactions had the most significant effect
on the market and counterparty default risk modules, with only a low impact on
operational risk and no impact at all on the other modules.

Page 138 of 153

© EIOPA 2011
However, as already mentioned, care should be taken in the interpretation of these
results since groups reported difficulties with calculating the adjustments for intra-
group transactions.

Market risk

The following table shows the impact of intra-group transactions on market risk,
broken down into the underlying sub-modules.

Table 31: Market risk intra-group transactions effects (ratio of sum of solo adjusted market
risk to sum of solo market risk)
10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Market risk 84% 99% 100% 100% 100% 92% 11% 141
Interest rate 100% 100% 100% 100% 100% 100% 16% 138
Equity 68% 95% 100% 100% 100% 80% 149% 134
Property 100% 100% 100% 100% 100% 99% 15% 114
Spread 100% 100% 100% 100% 100% 99% 4% 139
Currency 100% 100% 100% 100% 100% 97% 19% 113
Concentration 99.9% 100% 100% 100% 100% 75% 13% 121

On average, the impact of intra-group transactions was 8% of overall market risk. In

particular, the impact was material for concentration risk (25%) and equity risk
(20%), while it was practically negligible for spread, interest rate, currency and
property.

The high impact of intra-group transactions on concentration risk was linked to a

small number of mutual groups, whose mutual insurance undertakings own a high
share of proprietary companies. Concentration risk was therefore very high at solo
level, as the assets of these mutual undertakings are largely composed of these
participations. Due to consolidation, that concentration effect was eliminated at group
level.

Counterparty default risk

Table 32: Counterparty default risk intra-group transactions effects

10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Sum of solo adjusted
counterparty default risk
82% 99% 100% 100% 100% 62% 19% 138
to sum of solo
counterparty default risk

On average, the impact of intra-group transactions represented 38% of counterparty

default risk. One outlier group had a very high counterparty risk charge at the solo
level which related to an internal reinsurance special purpose vehicle.

Diversification deriving from the application of the solo standard formula at

group level

This section focuses on the effects of “real” diversification on the modules and sub-
modules of the SCR. As before, the ratio used below is between the SCR calculated
according to the consolidated method and the sum of solo adjusted SCRs.
Page 139 of 153
© EIOPA 2011
The following table shows the SCR modules for which the impact of “real”
diversification was significant.

Table 33: Real diversification by risk module

10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Market 58% 86% 95% 99% 101% 89% 22% 139
Health
81% 90% 99% 100% 100% 90% 37% 128
underwriting
Default 62% 90% 99% 100% 116% 84% 36% 135
Life
78% 89% 99% 100% 100% 92% 14% 115
underwriting
Non-life
79% 93% 100% 100% 102% 91% 145% 114
underwriting

Market risk

In terms of the sub-modules of market risk (see next table), interest, equity and
concentration risks were significantly impacted, with highly variable outcomes
between groups.

For some risks, “real” diversification appeared to be negative. This was especially
evident for currency risk. This result can partly be explained by two considerations:
firstly, groups included currency risks to holding companies when calculating the
consolidated SCR but not when summing the solo SCRs; secondly, the QIS5 currency
risks were calculated at group level on all own funds within the solo undertakings, and
not only on the solo surpluses, creating more currency risk at group level when the
assets of a solo undertaking were in a different currency to the currency of the group.

Table 34: Real diversification for market risk and its sub-modules
10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Market risk 59% 85% 95% 99% 101% 89% 22% 139
Interest
47% 65% 94% 100% 100% 74% 31% 130
rate
Equity 46% 91% 100% 100% 112% 87% 290% 130
Property 99% 100% 100% 100% 114% 103% 121% 111
Spread 96% 100% 100% 100% 107% 103% 18% 136
Currency 78% 100% 100% 100% 138% 130% 53% 108
Concentrati
41% 84% 100% 100% 115% 85% 189% 86
on

Non-life catastrophe risk

In the non-life underwriting risk module, the impact of “real” diversification is most
significant in the catastrophe sub-module which displayed diversification of 12%.

Page 140 of 153

© EIOPA 2011
Table 35: Real diversification for non-life catastrophe risk
10th 25th 50th 75th 90th Weighted Standard Sample
Percentile average deviation size
Non-life
catastrophe 76% 90% 99% 100% 101% 88% 200% 104
risk

The level of diversification can be explained by the way the sub-module is calculated
according to the technical specifications. When summing the solo adjusted SCRs for
non-life CAT risk, the most serious scenarios for each of the different undertakings are
taken on board, whereas when computing the group SCR only the most serious
scenario for the group will be kept. As a result, the calculation of the worst scenario at
group level is lower than the sum of the solo worst scenarios.

9.4. Availability constraints on group own funds

The own funds of the related undertakings included in a group cannot all be
considered available to cover the group SCR. For this reason groups have been asked
to consider possible restrictions on own funds items located in solo entities which
could prevent them being available to absorb losses elsewhere in the group.

For each related insurance undertaking, any non-available component of own funds in
excess of that undertaking‟s contribution to the group SCR should not be included in
group available own funds.

In addition, specific rules were set out for the treatment of minority interests: any
minority interests in the eligible own funds exceeding the SCR of the subsidiary
insurance undertaking should not be considered available for the group.

Table 36: Excess non-available own funds and minority interests

Percentage of
€bn
total own funds Sample size
Total excess non-available own funds
and minority interests 32.8 8% 109
Ring-fenced funds 5.4 21% 13
Other non-available own funds 16.8 5% 103
Minority interests 9.6 37% 45

109 groups have reported approximately €33bn of own funds which were not available
to cover the group SCR (including minority interests). This is equal to approximately
8% of those groups‟ total own funds. Responses from groups indicated that
restrictions on the availability of own funds were mainly related to surplus funds, ring-
fenced funds and equalisation reserves. In addition, a few groups and supervisors also
mentioned restrictions relating to own funds from entities located in non-EEA
countries.

Only thirteen groups reported ring-fenced funds. The fact that in contrast 218 ring-
fenced funds were reported at solo level suggests that many groups did not report
their ring-fenced funds, making it difficult to draw any conclusions.

As regards the non-availability of minority interests at group level, not all groups have
applied the limitations required in the technical specifications: 66% of groups
participating in QIS5 reported minority interests in their balance sheet, amounting to
Page 141 of 153
© EIOPA 2011
€26bn in total. However, only 45 groups reported non-available minority interests, of
€10bn altogether.

A few groups indicated that their minority interests were not material and so did not
make the required adjustments.

9.5. Other topics

9.5.1. Floor of the SCR

A group SCR floor applies when using the default method, equal to the sum of the
MCRs of the participating insurance undertakings and the proportional share of the
MCRs of the related insurance undertakings. On average, the group SCR floor is
42.3% of the group SCR. As a result, the floor has no impact on the group SCR.

9.5.2. Exchange of information within colleges of supervisors

Most group supervisors of large cross-border groups indicated either that they have
already shared their QIS5 results with the college (15 groups) or that they plan to do
so in early 2011 (19 groups). In general, the exchange of views and information on
QIS5 results and conclusions was deemed useful by group supervisors. In a few cases
the sharing of information with supervisors outside the EEA was considered of less
use. One EEA group supervisor mentioned that national legal restrictions on
confidentiality inhibited the sharing of QIS5 data related to the subsidiaries it
supervises.

It should be noted that the sharing of QIS5 submissions within colleges is one of the
targets in the EIOPA action plan for 2011 in relation to colleges of supervisors.

9.5.3. The potential effect of the diversification of the group risk margin

QIS5 aimed to test the potential effect of allowing diversification between entities
when calculating the risk margins of insurance groups. This was done by comparing a
default calculation (the sum of risk margins without diversification) with a calculation
of the group diversified risk margin. However, in most cases the latter calculation was
not provided, or did not give plausible results, so no direct assessment of the potential
effect of such diversification has been possible.

However, EIOPA has conducted an alternative assessment of the potential

diversification effect in the risk margin. For this assessment the potential
diversification in the risk margin has been approximated based on the diversification
effects of the group SCR. Just as in the solo level risk margin calculation, the SCR
needs to be recalculated assuming that only the unavoidable risk modules are
present.

Using the “recalculated” SCR the impact of diversification in the risk margin was
assessed to be between 13.5% and 16% of the overall risk margin. As a percentage
of the group SCR, the impact of diversification in the risk margin is between 3.8% and
4.8%.

Page 142 of 153

© EIOPA 2011
10. Practicability and preparedness

10.1. Preparedness

The vast majority of undertakings reported that they were not yet fully prepared for
Solvency II implementation, but that they expected to be by end 2012. The remainder
included both those already prepared and those who felt they might miss the
implementation deadline. However a few supervisors raised concerns that some
undertakings were underestimating what is required.

Some countries identified no particular trends in the degree of preparedness between

large and small undertakings, but others found that larger undertakings were
generally better prepared. Some countries also noted that undertakings within groups
tended to be better prepared, as they could draw on the group‟s expertise. One
country remarked that undertakings which had taken part in previous QIS exercises
had generally been better prepared.

A few countries noted that the complexity of the requirements was making
preparations challenging, and others noted that concurrent changes to accounting
standards and local regulations were also contributory factors. Several countries
raised concerns that any significant changes to the requirements prior to
implementation could cause problems or delays, and some observed that it was
challenging to assess preparedness when the requirements had not been finalised.

In terms of the key areas of focus for preparation, almost all countries cited ensuring
adequate quantity and/or quality of resource. The need for actuarial resources was
particularly highlighted, with risk management also being mentioned. However in the
context of this high demand, several countries raised concerns regarding the
availability of resource in the market. Some also noted a resultant dependency on
external consultants, especially among smaller undertakings.

Many countries also noted that training of existing resource would be an important
activity in undertakings‟ preparations, with a number of them stating this would
involve increasing awareness of the Solvency II requirements in the wider business.

Other areas that were commented on in almost all cases were improvements to data
quality or data management and, sometimes linked to the latter, changes to IT.

Several countries said the Pillar II and III requirements would be among the key areas
of preparation, some citing the ORSA in particular. One supervisor felt that
undertakings were less prepared for Pillar II than Pillar I. A number of countries also
said that further development of risk management functions, systems, or policies
would be important.

Undertakings in some countries said that conducting a gap analysis would be an

important step in their preparations, while others referred to the need to put a
detailed project plan in place.

A few countries cited the alignment of existing processes with the Solvency II
requirements as an important area, while others specifically referred to the need to
bring reporting into line. In a few countries some undertakings reported that changes

Page 143 of 153

© EIOPA 2011
to organisational structure were required. Some also said that they would have to
either strengthen their corporate governance or better align it with Solvency II.

Some countries noted that their undertakings needed support from their parent
groups in their preparations. Finally, a few countries cited working to better
understand the requirements as an important activity in their preparation.

10.2. Practicability
This section aims to draw together the various practicability concerns raised by
participants and supervisors, although many are covered in greater detail elsewhere in
this report.

10.2.1. Complexity

All countries commented on at least some areas of the QIS5 specifications as being
overly complex, burdensome and in some cases poorly understood by undertakings.
Some noted that this imposed additional costs on undertakings in terms of the
resources required for completion of the exercise. A few also commented that some
elements seemed disproportionately complex given their overall impact, in particular
the counterparty default risk sub-module and risk margin calculation.

The counterparty default risk sub-module was seen as overly complex by the vast
majority of countries, and a few reported that some undertakings omitted it entirely.
Respondents particularly referred to difficulties with the calculation of the risk-
mitigating impact of counterparties and the production of LGDs. The calculations for
reinsurance counterparties and the hypothetical SCR were also cited as causing
problems.

The risk margin calculation was another element widely regarded as very difficult to
complete without simplifications. Respondents commented in particular on the need to
produce future SCRs and the unavoidable market risk element.

The equivalent scenario approach was viewed as extremely complex by the vast
majority of countries, and in a significant number it was reported that it had not been
calculated by many, or in some countries any, participants. The SCR adjustment in
general was seen to be burdensome or of little added value, for deferred tax in
particular, and a number of countries reported undertakings leaving at least some
elements out.

Several countries felt that the adjustment for reinsurance, particularly non-
proportional reinsurance, posed a challenge, as did the recognition of reinsurance in
the CAT risk module and the valuation of reinsurance recoverables.

Other areas of complexity cited by a significant number of countries were the non-life
and health CAT risk sub-modules, the EPIFP calculation, and the inclusion of
embedded options and guarantees in contracts.

The look-through approach also drew comments from several countries, with
structured credit, collective investment schemes and investment funds cited as
products where this was particularly burdensome.

Page 144 of 153

© EIOPA 2011
The lapse risk module and especially the requirement to calculate on a policy-by-
policy basis was seen as complicated by a number of countries, as were contract
boundaries, the illiquidity premium and the illiquidity premium shock.

Several countries reported that it had at times been difficult to reconcile the QIS5
specifications with existing industry practices and accounting standards.

Other areas of particular complexity cited were:

The concentration risk sub-module.
Cash flow projections.
The premium provisions calculation.
The spread risk sub-module.
Inclusion and allocation of expenses.
The currency risk sub-module.
Undertaking-specific parameters.

10.2.2. Segmentation

Most countries also encountered some difficulties with segmenting business in the
manner required by the QIS specifications. This particularly affected health business
and the SLT/non-SLT split, with workers‟ compensation, personal accident insurance
and the split between income protection and medical expenses business also cited as
difficult areas.

Other areas mentioned were:

Segmentation of life business, particularly in the lapse and mortality/longevity
sub-modules.
Division of counterparties into type 1 and type 2 exposures.
Allocating the risk margin between LoBs.
Segmentation of motor business.
Segmentation of liability business.

10.2.3. Data requirements

There were also a number of areas where data requirements were seen as excessively
burdensome – these often coincided with areas seen as particularly complex.

Non-life and health CAT risk were the key areas where data requirements were seen
as cumbersome, particularly the man-made CAT shocks. The data requirements for
the counterparty default risk module were also widely found to be difficult to satisfy,
in particular in relation to reinsurance counterparties.

There were also some availability issues with the data needed for the look-through
approach, particularly in relation to structured credit, OEICs and investment funds.
Some countries also suggested there were weaknesses in the data which participants
used for the risk margin calculations.

Other data requirements found to be difficult to satisfy were:

For the calculation of best estimates.
For the adjustments for reinsurance.
Data on ratings for market risk.
For the calculation of future premiums.

Page 145 of 153

© EIOPA 2011
A number of countries noted that start-up undertakings struggled with elements
where historical data was required, such as in calculating technical provisions.

10.3. Guidance

There were a number of areas where members reported that the specifications were
not sufficiently clear or that further guidance would be appreciated.

Foremost among these was contract boundaries, commented on by a significant

number of countries. Several respondents also fed back that EPIFP and the illiquidity
premium could benefit from additional clarification.

Those were the only significant trends, but other areas where greater clarity was
called for were:
Segmentation of health business.
Ring-fenced funds.
The non-life CAT risk scenarios.
Application of the proportionality principle.
The adjustment for the loss absorbing capacity of deferred tax/technical
provisions.
The definition of strategic participations.
The single equivalent scenario.
The definitions of written or earned premium.
Lapse risk.
Definitions of counterparty types.

10.4. Technical Provisions simplifications

All of the proposed technical provisions simplifications were used by undertakings in

QIS5. The use of simplifications varied between countries but some were frequently
adopted by many undertakings across Europe. One reason for the great interest in
simplifications was that the calculation of technical provisions was generally found to
be complex in some areas, time-consuming, and demanding in terms of resources and
data. In some countries the use of simplifications was found to be most common
among small and medium-sized undertakings. Countries also indicated that in some
cases large undertakings also used simplifications.

Most countries agreed that the calculation of the risk margin was too complex and
observed that undertakings were often unable to carry out the full calculation.
Undertakings of all sizes adopted simplifications for the projection of the future SCR.
Countries strongly supported the use of simplifications for calculating the risk margin,
and felt this needed to remain a possibility in the future. See section 4.3.1 for further
details on the simplifications used.

In addition to the risk margin simplifications, the table below shows the other
frequently-used simplifications for technical provisions. We note that the simplification
that topped the list was the method based on expected claims ratio, used by
seventeen countries.

Page 146 of 153

© EIOPA 2011
Table 37: Most frequently-used technical provisions simplifications
Simplifications Number of
countries
used in
Premium provision - method based on expected claims ratio 17
Expenses and other charges 15
Financial options and guarantees 14
Biometric risk factors 13
Surrender option 13
Reinsurance recoverables – first simplification (duration-based formula) 13
Premium provision – method based on pro-rata of premiums 12
Outstanding claims provision – second simplification or sufficiently similar method 12
Other options and guarantees 10
IBNR claims provision – first simplification or sufficiently similar 10
Future discretionary benefits 9
Outstanding claims provision - first simplification or sufficiently similar 9
Investment guarantees 8
IBNR claims provision - second simplification or sufficiently similar 8
Reinsurance recoverable – second simplification (duration-based table) 8
Other life insurance simplifications 5

Countries also indicated that further development is needed to make the QIS5
simplifications easier to apply.

Some countries reported that more guidance on the application of the proportionality
principle is needed. Some of them asked for explanatory examples and a definition of
materiality thresholds.

Some other countries said that in general the application of the proportionality
principle seemed sufficiently clear for non-life, whereas for life business it had not
been used as often due to ambiguities in the technical specifications.

Other countries considered the QIS5 Technical Specifications to be sufficiently clear on

the application of the proportionality principle.

One country proposed a simplified method for the calculation of claims provisions in
health due to the lack of a long-tail distribution in the claims provisions.

Some countries expressed the view that the simplifications were not always
appropriate and felt that undertaking-specific simplifications should be allowed.

One country suggested that the simplification for options and guarantees is needed for
small undertakings.

Some countries noted that a simplification for EPIFP would be helpful for
undertakings.

Another country mentioned that for annuity business technical provisions a

deterministic approach could be taken to contracts where there is no policyholder
optionality, instead of a probability-weighted average of future cash flows approach.
The same country said that some undertakings had remarked that for some business
classified under life it would be very helpful, and appropriate for the business lines in
question, to use certain non-life simplifications.

Page 147 of 153

© EIOPA 2011
10.5. SCR simplifications

10.5.1. Existing simplifications

Use of simplifications

The simplifications were reasonably widely used and all simplifications were used by at
least some undertakings. By far the most adopted was the simplified approach offered
for counterparty default risk.

Graph 61: Use of Simplifications

600 50%

500
40%
Number reporting use

Proportion of BSCR
400
30%

300

20%
200

10%
100

0 0%
i ty

y
y

th
d

lt
e
lit
lit

au
a

ps

l
ev

ea
en
re

ta

bi

ef
La
a
Sp

ng
or

H
p

D
is

Ex
M

fe
Lo

Li
fe

Number of undertakings reporting use

Li

of the simplification

Proportion of the BSCR covered

where simplifications used

The above graph shows the number of undertaking which reported making use of the
different simplifications: this should be seen in the context that some supervisors
were of the view that the true number of undertakings making use of simplifications
might be some way higher than those reporting it. The graph also shows on average
what proportion of the BSCR was affected by the simplification in the cases where it
was used.

Where simplifications were used, they could sometimes affect a substantial proportion
of the SCR, particularly where they were adopted for mortality, longevity and health.

Page 148 of 153

© EIOPA 2011
Comments on simplifications

There was not a great deal of comment on the existing simplifications, indicating that
they were generally well received. As the most used, the counterparty default risk
simplifications were correspondingly the most commented on, though still only by
undertakings in a handful of countries, with a variety of concerns being raised:
that the simplification for calculating the risk-mitigating effect of reinsurance
counterparties failed to take into account diversification between perils and
lines of business;
that it did not take into account catastrophe risk mitigation;
that the allocation of risk-mitigating effect between counterparties was arbitrary
and didn‟t reflect the true risk position, although other participants felt this was
preferable to calculating the effect for each counterparty individually;
that no allowance was made for the number of reinsurers; and
that if the reinsurer for the coming year was not the same as in the previous
year, it would not reflect the risk appropriately.

Individual countries also raised concerns in relation to disability, longevity and lapse
simplifications: one comment on the latter was that it still required the calculation of
surrender strains on a policy-by-policy basis and was therefore still quite complex.

There were also a couple of comments on the captives simplifications: that for interest
rate risk, factors rather than just the simplified durations should be given, and that for
shocks on technical provisions, durations should be given.

Comments on proportionality principle

A considerable number of countries felt that further guidance on the proportionality

principle and the use of simplifications would be useful. Some felt this was necessary
in order to ensure simplifications were not used inappropriately, although others
emphasised that this had to be balanced with sufficient flexibility in the criteria to
ensure that simplifications could be used where needed.

A few countries noted that the criteria for using the simplifications could be
paradoxical, in that in some cases you could not demonstrate that the criteria were
met without performing the calculation that the simplification was intended to
circumvent. Another comment suggested that the criteria be based on the relative
impact of the module on the SCR rather than the size of the undertaking.

10.5.2. Additional simplifications suggested

There were a number of areas in which undertakings requested or suggested
additional simplifications, often to elements of the standard formula which were seen
to be particularly complex. Some supervisors noted that in some of these cases what
was really needed was a reduction in the complexity of the standard formula, rather
than additional simplifications being made available, expressing their concern that
offering more choices to the standard formula might impair the comparability of
results.

Counterparty default risk

Although the existing simplifications for counterparty default risk were relatively
widely used, this was nonetheless cited by most countries as a key area where

Page 149 of 153

© EIOPA 2011
additional simplifications would be beneficial, with some countries suggesting that the
current simplifications were still too sophisticated.

In particular it was felt that calculating the risk-mitigating effect of counterparties)

could benefit from simplifications or a simplified standard approach. Suggestions
were:
That the risk-mitigating effect be set based on the SCR of the reinsurance;
Instead of recalculating SCR for each counterparty, to allow VaR values for
derivatives and something similar for reinsurers;
To calculate XL treaty reinsurance recoverables as the average recuperation of
the treaty to the reinsured;
That the total risk mitigation be calculated and allocated to counterparties
based on their share of recoverables;
That the risk-mitigating effect be set as a flat percentage of LGD, although this
was not supported by the supervisor;
To use the national accounting value for LGD where the exposure was not
material, or to use balance sheet values throughout; and
To group counterparties by rating.

In addition, undertakings in some countries offered ideas on overhauling the overall

approach of the counterparty risk calculation. Suggestions were:
To replace the variance term with a factor for each rating;
To reinstate the formula from QIS3; and
To use factors based on the asset value of the risk mitigant, the credit quality
of the counterparty, and the length and type of exposure (using higher factors
where the counterparty has a risk-mitigating effect).

Market risk

Look-through approach

A significant number of countries saw scope for simplifications to the look-through

approach used in the market risk module, particularly for investments in unit-linked
funds (with some noting that most market risk in relation to these liabilities would be
borne by the insured, and so it was not a material area). Simplifications proposed for
unit-linked funds were:
To use the asset type split by the fund‟s asset allocation or investment
mandate;
To use approximations of asset allocations, currencies, ratings and durations of
investments; and
To assume all assets are equities and make a high-level currency split.
The other key area of comment was structured credit, with a couple of participants
proposing that the direct approach (rather than look-through) be used for these
assets: supervisors were generally less supportive of this idea.

Spread risk

Spread risk also drew a number of comments. One country suggested a factor-based
approach, which should take into account the specificities of structured credit.
Undertakings in another proposed that the split by ratings should not be mandatory.
Finally, one other proposed simplification was to stress the difference between the
„current value‟ and a value assuming the implicit credit spread (the difference between
the internal rate of return and the current yield), plus a fixed percentage.

Page 150 of 153

© EIOPA 2011
A few countries noted the duration of bonds as an area where a simplification would
be appreciated (we note that there were also several questions in the QIS5 Q&A on
how to properly account for the duration of bonds and other securities) and suggested
that proxies be allowed, at least for floating rate bonds. Another country reported that
in some cases spot rates had been used for floating rate notes, rather than forward
rates.

Interest rate risk

A number of simplifications were also proposed in relation to interest rate risk.

However this was not highlighted by many as a key area and at least one supervisor
expressed the view that this sub-module was well within the capabilities of most
undertakings.

Most proposals involved modified duration approaches rather than actual cash flow
projections, some suggesting this only for unit-linked or externally managed
investments or short-term assets. One country reported a participant applying a fixed
rate to variable rate mortgages, and another suggested that the indexation of bonds
be taken into account.

Currency risk

A few simplifications were suggested in relation to currency risk, including the

aggregation of minor currencies into others, and a suggestion to conduct a single
economically equivalent stress rather than stressing each currency individually.

Other

Other areas of market risk where it was suggested that simplifications might be useful
included the application of the equity risk sub-module to convertible bonds, and that it
should be possible to group assets by rating, duration or similar where this was
proportionate.

Non-life underwriting risk

Non-life catastrophe

Another area that a number of countries cited as in need of simplifications was non-
life catastrophe risk, with the CRESTA zone and 150/300m radius data requirements
particularly noted as areas where this could be considered. A few countries suggested
that national catastrophe scenarios should be developed and then split by market
share. Another suggested that concentration scenarios should be based on postcodes
rather than radii.

Other

It was reported that in relation to premium and reserve risk, in some cases premiums
written were all considered to be already paid, even where this was not the case.
There was also a suggestion that the non-life lapse risk should be calculated
separately for annual and multi-year contracts.

Page 151 of 153

© EIOPA 2011
Health catastrophe risk

There were a few suggestions from undertakings around health catastrophe, along the
same lines as for non-life catastrophe:
That the most granular level of postcode be used instead of a set radius;
That a factor-based approach be used;
That the three scenarios be replaced by a single catastrophe event that applies
to all health insurance obligations;
That minor countries could be aggregated together; and
That a similar approach be used as with life catastrophe.

Life underwriting risk

Lapse risk

Several countries felt that simplifications could be introduced for lapse risk,
particularly around the calculation of surrender strains and the policy-by-policy
approach. Simplifications suggested were:
Allowing the use of model points;
Calculating the net positive and negative lapses rather than splitting by policy;
Calculating lapse up and lapse down on all policies rather than working out the
surrender strains; and
Calculating surrender strain at product rather than policy level.

There was also a suggested simplification for mass lapse: that an expense stress be
calculated instead using the increased expense costs for the remaining business if
there were 30% fewer policies in place.

Other

One country suggested that disability could be calculated as a single shock instead of
year by year. In a few countries undertakings suggested simplifications for the life
underwriting sub-modules in general, such as the use of a zero floor at contract level
for mortality and longevity, the removal of per-policy capping, and that the stresses
could be calculated at product rather than policy level. There was also an undertaking
which suggested the mortality/longevity split could be done by carrying out both
stresses and applying the most onerous.

Other

A few countries suggested there should be simplifications available for small sub-
portfolios, with a couple proposing that they could be aggregated with or
approximated by a larger portfolio. One country suggested that non-life guarantees
not be unbundled from life contracts.

In one country it was suggested that the calculation of geographical diversification

could be omitted where this was not material. In another country it was suggested
that operational risk could be calculated as a percentage of operating expenses.

Finally, it was proposed that where calculation at insured/contract level was easier
than at group policy level, this should always be available as an option.

Page 152 of 153

© EIOPA 2011
Adjustment for non-proportional reinsurance

Another area where simplifications were felt to be needed by several countries was
the adjustment for non-proportional reinsurance. One suggested simplification was
that the premium risk factors should be reduced by a factor based on non-
proportional reinsurance claims recoveries divided by net claims (pre recoveries),
averaged over 3-5 years. Another suggestion was that this should always be done as
in the non-life catastrophe sub-module.

Adjustment for the loss absorbing capacity of technical provisions and deferred taxes

A number of countries commented that simplifications would be useful for the loss
absorbing capacity of technical provisions, and usually suggested that deterministic
proxy approaches be allowed.

A couple felt that this would be helpful for the deferred taxes adjustment as well,
suggesting the use of an undertaking-specific average tax rate, or that this be
calculated as the difference in the change in NAV for all risks on a pre and post-tax
basis.

Page 153 of 153

© EIOPA 2011