First article

3.1.1. Explanatory text
Design and structure
3.1. A number of the life underwriting risk stresses are based on a delta-
NAV (change in value of assets minus liabilities) approach. The
change in net asset value should be based on a balance sheet that
does not include the risk margin of the technical provisions. This
approach is based on the assumption that the risk margin does not
change materially under the scenario stress. This simplification is
made to avoid a circular definition of the SCR since the size of the risk
margin depends on the SCR.
3.2. Furthermore, where a delta-NAV approach is used, the revaluation of
technical provisions should allow for any relevant adverse changes in
option take-up behaviour of policyholders in this scenario.
Calibration
3.3. The calibration of the life underwriting parameters should capture
changes in the level, trend and volatility of the parameter. However,
for QIS 3, it was decided to reduce the complexity of the design of
the underwriting risk module by maintaining the level and trend risk
components only. It is assumed that the volatility risk component is
implicitly covered by the level, trend and catastrophe risk
components. This is considered to be acceptable since, for QIS2, the
volatility risk proved to be considerably lower than the trend risk.
CEIOPS therefore proposes to retain this approach.
3.1.2. CEIOPS’ advice
General considerations
3.4. The change in net asset value shall be based on a balance sheet that does
not include the risk margin of the technical provisions.
3.5. The revaluation should allow for any relevant adverse changes in option
take-up behaviour of policyholders in this scenario.
3.6. The calibration of the life underwriting parameters shall capture changes in
the level and trend of the parameters only. It is assumed that the volatility
risk component is implicitly covered by the level, trend and catastrophe
risk components.
7/51
3.2 Mortality risk
3.2.1. Explanatory text
Introduction
3.7. Mortality risk is associated with (re)insurance obligations (such as term
assurance or endowment policies) where a (re)insurance undertaking
guarantees to make a single or recurring series of payments in the event
of the death of the policyholder during the policy term.
3.8. It is applicable for (re)insurance obligations contingent on mortality risk
i.e. where the amount currently payable on death exceeds the technical
provisions held and, as a result, an increase in mortality rates is likely to
lead to an increase in the technical provisions.
3.9. The capital charge for mortality risk is intended to reflect the uncertainty
in mortality parameters as a result of changes in the level, trend and
volatility of mortality rates and capture the risk that more policyholders
than anticipated die during the policy term.
3.10. This risk is normally captured by increasing the mortality rates either by a
fixed amount or by a proportion of the base mortality rates. The
calibration (of the increase) should capture the impact of each of the
above factors (level, trend and volatility).
Mortality risk in QIS4
3.11. The QIS4 approach to the SCR standard formula included a mortality risk
sub-module in the life underwriting risk module (section TS.XI.B of the
QIS4 Technical Specifications (MARKT/2505/08)). The calculation of the
capital requirement for mortality risk was a scenario based stress. The
scenario tested was a permanent 10% increase in mortality rates.
3.12. QIS4 feedback from several Member States suggested that a gradual
change to inception rates and trends would be more appropriate than a
one-off shock for biometric risks.
3.13. QIS4 feedback on the calibration of the mortality stress was varied. Some
undertaking felt that the calibration was too strong and without sufficient
granularity whereas other undertakings thought that the calibration was
below the 99.5th percentile.
3.14. QIS4 also tested alternative approaches for dealing with (re)insurance
obligations which provide benefits on both death and survival. The first
option proposed that where the death and survival benefits are contingent
on the life of the same insured person(s), the obligation should not be
unbundled. Under the second option, all contracts were unbundled into
two separate components: one contingent on the death and other
contingent on the survival of the insured person(s). Only the former
component was taken into account for the application of the mortality
scenario.
8/51
3.15. Feedback from QIS4 indicated that the vast majority of (re)insurance
undertakings chose not to unbundle the obligations (option one). The
practical difficulty in unbundling obligations was cited as the main reason
for choosing this option. Undertakings in one Member State also noted
that this (option one) was consistent with IFRS classifications. Where
supervisors offered views, they generally agreed with undertakings.
However one Member State argued that more analysis would be
necessary before deciding on the most appropriate option.
Calculation of the capital requirement
3.16. QIS4 participants suggested that a gradual change to inception rates and
trends would be more appropriate than a one-off shock for biometric
risks. However CEIOPS has considered this proposal (see in particular
discussion under longevity risk below) and has concluded that a one-off
shock is more appropriate in the context of the standard formula.
3.17. The capital requirement should therefore be calculated as the change in
net asset value (assets minus liabilities) following a permanent increase
in mortality rates of x%.
Calibration of mortality stress
3.18. The basis for the QIS4 calibration of the mortality risk stress is described
in the CEIOPS paper “QIS3 Calibration of underwriting risk, market risk
and MCR”. This paper is available from the CEIOPS website3.
3.19. As mentioned above, QIS4 feedback on the calibration of the mortality
stress was varied. However an analysis of the mortality stress parameters
provided by firms using internal models indicated that the standard
formula parameter was relatively low. Based on a sample size of 21
internal model, the median stress was 22%, with an inter quartile range
of 13% to 29%. This is significantly higher than the standard formula
calibration of 10%.
3.20. CEIOPS therefore proposes to amend the calibration of the mortality
stress to a permanent increase in mortality rates of 15%.
Unbundling of (re)insurance obligations
3.21. Where (re)insurance obligations provide benefits both in case of death
and survival and the death and survival benefits are contingent on the life
of the same insured person(s), these obligations should not be
unbundled. For these contracts the mortality scenario should be applied
fully allowing for the netting effect provided by the ‘natural’ hedge
between the death benefits component and the survival benefits
component (note that a floor of zero applies at the level of contract if the
net result of the scenario is favourable to the (re)insurer).
3.22. Where model points are used for the purposes of calculating the technical
provisions and the grouping of the data captures appropriately the
mortality risk of the portfolio, each model points can be considered to
3 http://www.ceiops.eu/media/files/consultations/QIS/QIS3/QIS3CalibrationPapers.pdf
9/51
represent a single insured person for the purposes of applying the above
advice.
3.2.2. CEIOPS’ advice
Mortality risk
3.23. The mortality risk sub-module is applicable for (re)insurance obligations
contingent on mortality risk i.e. where the amount currently payable on
death exceeds the technical provisions held and, as a result, an increase in
mortality rates leads to an increase in the technical provisions.
3.24. The calculation of the capital requirement for mortality risk shall be a
scenario based stress.
3.25. The capital requirement shall be calculated as the change in net asset
value (assets minus liabilities) following a permanent increase in mortality
rates of 15%.
3.46. Where (re)insurance obligations provide benefits both in case of death and
survival and the death and survival benefits are contingent on the life of
the same insured person(s), these obligations should not be unbundled.
For these contracts the mortality scenario should be applied fully allowing
for the netting effect provided by the ‘natural’ hedge between the death
benefits component and the survival benefits component (note that a floor
of zero applies at the level of contract if the net result of the scenario is
favourable to the (re)insurer).
3.47. Where model points are used for the purposes of calculating the technical
provisions and the grouping of the data captures appropriately the
mortality risk of the portfolio, each model points can be considered to
represent a single insured person for the purposes of applying the above
advice.
3.3. Longevity risk
3.3.1. Explanatory text
Introduction
3.26. Longevity risk is associated with (re)insurance obligations (such as
annuities) where a (re)insurance undertaking guarantees to make
recurring series of payments until the death of the policyholder and where
a decrease in mortality rates leads to an increase in the technical
provisions, or with (re)insurance obligations (such as pure endowments)
where a (re)insurance undertaking guarantees to make a single payment
in the event of the survival of the policyholder for the duration of the
policy term.
3.27. It is applicable for (re)insurance obligations contingent on longevity risk
i.e. where there is no death benefit or the amount currently payable on
10/51
death is less than the technical provisions held and, as a result, a decrease
in mortality rates is likely to lead to an increase in the technical provisions.
3.28. The risk that a policyholder lives longer than anticipated is longevity risk.
Longevity risk is particularly significant as a result of an increasing life
expectancy among policyholders in most developed countries.
3.29. The capital charge for longevity risk is intended to reflect the uncertainty
in mortality parameters as a result of changes in the level, trend and
volatility of mortality rates and capture the risk of policyholders living
longer than anticipated.
3.30. This risk may be captured in a number of different ways: a simple
approach of a reduction in base mortality rates, a more realistic approach
of using improvement factors which leads to a two dimensional mortality
table, or a combination of these two approaches. In any event, the
calibration (of the increase) should capture the impact of each of the
above factors (level, trend and volatility).
Longevity risk in QIS4
3.31. The QIS4 approach to the SCR standard formula included a longevity risk
sub-module in the life underwriting risk module (section TS.XI.C of the
QIS4 Technical Specifications (MARKT/2505/08)). The calculation of the
capital requirement for longevity risk was a scenario based stress. The
scenario tested was a permanent 25% decrease in mortality rates.
3.32. QIS4 feedback from several Member States suggested that a gradual
change to inception rates and trends would be more appropriate than a
one-off shock for biometric risks.
3.33. With regard to the calibration of the longevity stress, several undertakings
argued for an age and duration dependent treatment of longevity,
reinforcing more general comments that a one-off shock is not the most
appropriate form of stress for biometric risks. An improvement of x% per
annum (over base mortality) was suggested as an alternative by one
respondent.
3.34. Some undertakings felt the longevity shock was too conservative.
Calculation of the capital requirement
3.35. QIS4 participants suggested that a gradual change to inception rates and
trends would be more appropriate than a one-off shock for biometric risks.
For example, one respondent suggested that an improvement of x% per
annum (over base mortality) could be used as an alternative.
3.36. Subsequent to QIS4, an analysis by UNESPA proposed an alternative
structure to the longevity shock which depended on age and duration.
3.37. CEIOPS has considered the above mentioned proposals but has concluded
that a one-off shock to longevity is more appropriate for the purposes of
the standard formula for the following reasons:
11/51
• It is more straightforward to apply
• With respect to differentiating by duration, CEIOPS’ investigations (see
Appendix B to this paper) indicate that shocks for different durations
are small and are not monotone.
• With respect to differentiating by age, portfolios of (re)insurance
obligations for which longevity risk is applicable are generally heavily
weighted in favour of older age groups.
• We do not believe that there is sufficient reliable data to calibrate at a
more granular level
3.38. The capital requirement should therefore be calculated as the change in
net asset value (assets minus liabilities) following a permanent decrease in
mortality rates of x%.
Calibration of longevity stress
3.39. The basis for the QIS4 calibration of the longevity risk stress is described
in the CEIOPS paper “QIS3 Calibration of underwriting risk, market risk
and MCR”.4
3.40. Subsequent to QIS4, an investigation has been carried out by the Polish
FSA which analysed the mortality data for nine countries indicated based
both on historic improvements and a stochastic model of future mortality
improvements.
3.41. The results of this analysis indicated that, on average (across the nine
countries for which data was analysed), historic improvements in mortality
rates over 15 years from 1992 to 2006 were higher than 25%. Although
the results of the stochastic model of future mortality improvements may
imply a lower stress, CEIOPS has attached more weight to the analysis of
historic improvements because of the significant uncertainty inherent in
modelling mortality.
3.42. Furthermore feedback from internal model firms as part of QIS4 indicates
that the median stress was 25%.
3.43. CEIOPS therefore proposes to maintain the QIS4 calibration of the
longevity risk stress i.e. the stress shall be based on a permanent 25%
decrease in mortality rates.
Unbundling of (re)insurance obligations
3.44. Where (re)insurance obligations provide benefits both in case of death and
survival and the death and survival benefits are contingent on the life of
the same insured person(s), these obligations should not be unbundled.
For these contracts the longevity scenario should be applied fully allowing
for the netting effect provided by the ‘natural’ hedge between the death
benefits component and the survival benefits component (note that a floor
4 http://www.ceiops.eu/media/files/consultations/QIS/QIS3/QIS3CalibrationPapers.pdf
12/51
of zero applies at the level of contract if the net result of the scenario is
favourable to the (re)insurer).
3.45. Where model points are used for the purposes of calculating the technical
provisions and the grouping of the data captures appropriately the
longevity risk of the portfolio, each model points can be considered to
represent a single insured person for the purposes of applying the above
advice.
3.3.2. CEIOPS’ advice
Longevity risk
3.48. The longevity risk sub-module is applicable for (re)insurance obligations
contingent on longevity risk i.e. i.e. where there is no death benefit or the
amount currently payable on death is less than the technical provisions
held and, as a result, a decrease in mortality rates is likely to lead to an
increase in the technical provisions.
3.49. The calculation of the capital requirement for longevity risk shall be a
scenario based stress.
3.50. The capital requirement shall be calculated as the change in net asset
value (assets minus liabilities) following a permanent decrease in mortality
rates of 25%.
3.51. Where (re)insurance obligations provide benefits both in case of death and
survival and the death and survival benefits are contingent on the life of
the same insured person(s), these obligations should not be unbundled.
For these contracts the longevity scenario should be applied fully allowing
for the netting effect provided by the ‘natural’ hedge between the death
benefits component and the survival benefits component (note that a floor
of zero applies at the level of contract if the net result of the scenario is
favourable to the (re)insurer).
3.52. Where model points are used for the purposes of calculating the technical
provisions and the grouping of the data captures appropriately the
longevity risk of the portfolio, each model points can be considered to
represent a single insured person for the purposes of applying the above
advice.
3.4. Disability-morbidity risk
3.4.1. Explanatory text
Introduction
3.51. Morbidity or disability risk is associated with all types of insurance
compensating or reimbursing losses (e.g. loss of income) caused by
illness, accident or disability (income insurance), or medical expenses due
to illness, accident or disability (medical insurance).
13/51
3.52. It is applicable for (re)insurance obligations contingent on a definition of
disability. However CEIOPS expects that the majority of (re)insurance
obligations for which disability-morbidity risk is applicable will be covered
by the health module rather than by the life underwriting module. This
sub-module of the life underwriting risk module is therefore likely to be
applicable only in cases where contracts cannot be unbundled.
3.53. The capital charge for morbidity or disability risk is intended to reflect the
uncertainty in morbidity and disability parameters as a result of changes in
the level, trend and volatility of disability, sickness and morbidity rates
and capture the risk that more policyholders than anticipated are
diagnosed with the diseases covered or are or unable to work as a result
of sickness or disability during the policy term.
3.54. The (re)insurance obligations may be structured such that, upon the
diagnosis of a disease or the policyholder being unable to work as a result
of sickness or disability, recurring payments are triggered. These
payments may continue until the expiry of some defined period of time or
until either the recovery or death of the policyholder. In the latter case,
the (re)insurance undertaking is also exposed to the risk that the
policyholders receives the payments for longer than anticipated i.e. that
claim termination rates are lower than anticipated (recovery risk).
3.55. Morbidity and disability risk is normally captured by increasing the claim
inception rate either by a fixed amount or by a proportion of the base
inception rates and, where applicable, reducing the claim termination
rates. The calibration (of the increase) should capture the impact of each
of the above factors (level, trend and volatility).
Morbidity and disability risk in QIS4
3.56. The QIS4 approach to the SCR standard formula included a morbidity and
disability risk sub-module in the life underwriting risk module (section
TS.XI.B of the QIS4 Technical Specifications (MARKT/2505/08)). The
calculation of the capital requirement for morbidity and disability risk was
a scenario based stress. The scenario tested was an increase of 35% to
“disability rates” for the first year followed by a 25% increase in “disability
rates” for all subsequent years.
3.57. An alternative scenario was also proposed by the UK under which the
capital charges for critical illness, income protection and long term care
obligations were calculated separately and there was an additional capital
charge in respect of recovery risk.
3.58. There were a number of comments from QIS4 participants on the general
methodology of the morbidity and disability stress:
• One respondent argued that recovery rates should be taken into
account.
• There was some confusion over the treatment of disability in terms of
catastrophe risk.
14/51
• Support for the UK alternative approach was noted by one Member
State.
3.59. With respect to the calibration of the morbidity and disability stress, some
(re)insurance undertakings commented that the calibration was too
strong.
Calculation of the capital requirement
3.60. As described above, there are two aspects to morbidity/disability risk:
• The risk that the number of claims are greater than anticipated
• The risk that the duration of the claim is higher than anticipated
The second risk is only applicable for (re)insurance obligations where
benefits consist of recurring payments which continue until either the
recovery or death of the policyholder.
3.61. Therefore the capital requirement should be calculated as:
• The change in net asset value (assets minus liabilities) following an
increase of x1% in morbidity/disability inception rates for the first year
followed by an increase of x2% in morbidity/disability inception rates
for all subsequent years.
• Plus, where applicable, the change in net asset value (assets minus
liabilities) following a permanent decrease of y% in morbidity/disability
recovery rates
Calibration of morbidity and disability stress
3.62. The basis for the QIS4 calibration of the morbidity-disability risk stress is
described in the CEIOPS paper “QIS3 Calibration of underwriting risk,
market risk and MCR”. This paper is available from the CEIOPS website.
3.63. Subsequent to QIS4, an investigation by the Swedish FSA indicated that
an increase of 50% in morbidity/disability inception rates for the first year
would be more appropriate.
3.64. This investigation also suggested that the appropriate calibration of the
decrease in morbidity/disability recovery rates was 20%.
3.65. The results of the investigation by the Swedish FSA are explained further
in Appendix A.
3.66. In addition, the UK Actuarial Profession Healthcare Reserving Working
Party has undertaken a survey which investigated the levels of 1 in 200
year morbidity stresses used by the major UK life insurance firms.5
3.67. The range of stress used by the major UK life insurers for income
protection business averaged 27% for inception rates and 15% for
5 http://www.actuaries.org.uk/__data/assets/pdf_file/0006/136707/reserving_survey.pdf
15/51
termination rates. For critical illness, morbidity margins, intended to
represent a 99.5% confidence over 1 year, averaged around 40%.
3.68. Furthermore, on average, the average morbidity margins for statutory
reserving for critical illness and income protection (both inceptions and
terminations) were about 20%. The margins in a statutory reserving basis
are partly to allow for adverse deviations of the inception and termination
rates used in the pricing. As such, a 1 in 200 stress should be at least
greater than these margins as these margins are not normally set at the
same level as a 1 in 200 year scenario.
3.69. Looking at the results of this survey in conjunction with the results of the
investigation by the Swedish FSA, we would propose the following
calibration of the disability-morbidity stress:
• The change in net asset value (assets minus liabilities) following an
increase of 50% in morbidity/disability inception rates for the first year
followed by an increase of 25% in morbidity/disability inception rates
for all subsequent years.
• Plus, where applicable, the change in net asset value (assets minus
liabilities) following a permanent decrease of 20% in
morbidity/disability recovery rates. This should be applied together
with the above increase in inception rates i.e. it is a combined stress.
3.4.2. CEIOPS’ advice
Morbidity-disability risk
3.70. The morbidity-disability risk sub-module is applicable for (re)insurance
obligations contingent on a definition of disability.
3.71. The calculation of the capital requirement for disability risk shall be a
scenario based stress.
3.72. The capital requirement shall be calculated as the change in net asset
value (assets minus liabilities) following:
• An increase of 50% in morbidity/disability inception rates for the
first year followed by an increase of 25% in morbidity/disability
inception rates for all subsequent years.
• Plus, where applicable, a permanent decrease of 20% in
morbidity/disability recovery rates.
16/51
3.5 Life expense risk
3.5.1. Explanatory text
Introduction
3.73. Expense risk arises from the variation in the expenses incurred in servicing
insurance or reinsurance contracts.
3.74. It is likely to be applicable for all (re)insurance obligations.
3.75. The capital charge for expense risk is intended to reflect the uncertainty in
expense parameters as a result of changes in the level, trend or volatility
the expenses incurred.
3.76. This risk is normally captured by increasing expected future expenses by a
fixed proportion, increasing expected future expense inflation or a
combination of both.
Expense risk in QIS4
3.77. The QIS4 approach to the SCR standard formula included an expense risk
sub-module in the life underwriting risk module (section TS.XI.F of the
QIS4 Technical Specifications (MARKT/2505/08)). The calculation of the
capital requirement for expense risk was a scenario based stress. The
scenario tested was:
• An increase of 10% in future expenses compared to best estimate
anticipations,
• An increase of 1% per annum of the expense inflation rate compared
to anticipations
For policies with adjustable loadings6, 75% of these additional expenses
can be recovered from year 2 onwards by increasing the charges payable
by policyholders.
3.78. There was a range of opinions with regard to the calibration of the
expense risk as a result of which no useful conclusion could be drawn.
Calculation of the capital requirement
3.79. QIS4 participants did not raise any significant issues with the design and
structure of this module and CEIOPS has therefore concluded that the
approach adopted in QIS4 is appropriate.
3.80. The capital requirement should therefore be calculated as the change in
net asset value (assets minus liabilities) following:
6 Policies with adjustable loadings are those for which expense loadings or charges may be adjusted within the next
12 months.
17/51
• An increase of x% in future expenses compared to best estimate
anticipations,
• An increase of y% per annum of the expense inflation rate compared
to anticipations
3.81. However CEIOPS does not intend to retain the specific reference to policies
with adjustable loadings. This is because any future change to charges
payable by policyholders is, in essence, a management action and should
thus be considered in light of CEIOPS’ advice on management actions
rather than specified by CEIOPS.
Calibration of expense stress
3.82. The basis for the QIS4 calibration of the expense risk stress is described in
the CEIOPS paper “QIS3 Calibration of underwriting risk, market risk and
MCR”. This paper is available from the CEIOPS website.
3.83. As mentioned above, QIS4 feedback on the calibration of the expense
stress was varied. However the expense risk capital charge from the
internal model tended to be, for many undertakings, in line with the
standard formula. The median ratio was equal to 100% and the inter
quartile range was 85% to 166%.
3.84. CEIOPS therefore proposes to maintain the QIS4 calibration of the
expense risk stress i.e. the stress shall be based on:
• An increase of 10% in future expenses compared to best estimate
anticipations,
• An increase of 1% per annum of the expense inflation rate compared
to anticipations