1
Questions which should be answered using the EXCEL spreadsheet are formatted in Italic letters.
Please use this word document only for answering the other questions or indicating additional opinions that could not be answered through the use of the given spreadsheet format.
<QIMx
In order to ease the treatment of your submission by the FSB , please reply to the question filling the grey shaded area between the provided beginning and ending tags <Qx> and </Qx>, inserting as many lines as necessary. Please also give any other information that doesn’t follow the structure of the predefined questions at the beginning between the <Q0> and </Q0> tags.
</QIMx
(Re)insurer name:
<QIMP>
</QIMP>
(Re)insurer number:
QIMR>
</QIMR>
1/35
FSB 2011
Internal models
IM.A Introduction and background
In order to allow an assessment of the impact of the use of internal models under SAM, (re)insurers should answer the questions listed in this questionnaire. Furthermore, the answers to these questions will help to assess the progress of internal modelling and to prepare the FSB and the insurance industry to the approval process of internal models.
The goals of the firstSouth African Quantitative Impact Study (SA QIS1) for internal models are:
(a)to collect reliable and comparable quantitative and qualitative data from partial and full internal models that are currently used by (re)insurers for assessing their capital needs. . This data will assist the FSB in conducting a range of statistical analyses and then providing a possible update of the calibration of the standard formula and its likely impacts on the SAM regime;
(b)to help (re)insurers in an assessment of how they are advanced in fulfilling the conditions and identify potential gaps between the current stage of their models and the requirements;
(c)to collect general information from all (re)insurers to assess the current and potential future levels of applications for full and partial internal models to enable the FSB to prepare itself and to provide an estimate of the costs related to the application for (re)insurers and the FSB.
To achieve the first goal, (re)insurers will have to assess the quality and comparability of the data against high level principles. Therefore (re)insurers should concentrate on comparing the results and the modelling aspects of the standard formula with those derived from their internal models. A key area to address in this context is the data that has been used when (re)insurers calibrate their models. It is important also to understand the differences in assumptions and definitions between those underlying existing models in (re)insurers and those anticipated under SAM[1].
To this end, and to the extent that this is practicable, the estimates derived from internal models should be compatible with the overall calibration objectives for the standard formula (i.e. a VaR 99.5% confidence level for the variation of basic own funds over a one year time horizon).
The importance of qualitative issues is highlighted by the second goal. The FSB is interested in the assessment how the internal models at the current stage comply with the standard requirements.
Finally, the SA QIS1 exercise should also serve as a tentative mapping exercise of the current development stage of internal models used by (re)insurers, and to indicate to what extent they plan to use internal models for calculating their SCR or use partial internal models to calculate modules of the SCR or in respect of some or all modules for some but not all of their business units. To gain an accurate picture of current developments, internal models referred to in this section should be understood as comprising those that include any risk management analysis to quantify risks and to help to assess the economic capital needed to meet those risks.
It should be noted that a disaggregation of the output from internal models to the level of granularity of the standard formula may not be feasible for all internal models and internal lines of business may not be fully compatible with those used in the QIS. However, internal estimates for capital corresponding to main risk classes and the overall SCR is especially welcomed (both solo-entity and group results). Similarly, more granular results (risk sub-classes) and capital requirements for lines of business or modules of the internal model that are different from the standard formula modules are also welcomed.
The integration technique used by (re)insurers to integrate a partial internal model into the Solvency Capital Requirement standard formula should result in a Solvency Capital Requirement for the (re)insurer which complies with the principles included in point 4.
In order to fully integrate the partial internal model into the standard formula for the purpose of calculating the Solvency Capital Requirement, (re)insurers should use the correlation matrices of the standard formula set out in these technical specifications unless there is strong evidence that:
–those correlation matrices are not possible to be used, having regard to the structure of the (re)insurer’s partial internal model; or
–The requirements set out in the previous points would not be met.
In the event that correlation matrices are not possible to be used or that the requirements set out would not be met, (re)insurers should use an alternative integration technique provided that the requirements are met.
Some specific data requests on internal models will be also included in the spreadsheet (e.g. correlations and diversification effects at different levels, risk mitigation effects, scenario parameters, volatility parameters). Moreover, description of the original calibration (i.e. before any recalibration) for the main risk categories should be disclosed (e.g. risk measure, confidence level, time horizon).
Groups are required to provide the results obtained by their internal model implemented for the whole group. (Re)insurers should also provide the amounts of diversification effects obtained for each level of aggregation of risks.
Guidance how to fill in the questionnaire:
Under the SAM regime, groups may apply for an internal model which would be used to assess both the group SCR and the solo SCR for (re)insurers within the group. In this paper, this type of internal model will be referred to as a group internal model.
The qualitative questionnaire for internal models is organised in 4 different sections (see table below).
(1)(Re)insurers
a) which are not part of a group and which currently use or intend to use an internal model or,
b) which are part of a group, and intending to use an internal model for the solo SCR calculation which is not a group internal model,
should answer to the questions of the sections IM.B, IM.C, but not to the questions in the section IM.D and in the group section.
(2)(Re)insurers being part of a group and intending to use a group internal model for the solo SCR calculation are requested to answer to the questions in the sections IM.B, IM.C, IM.D but not in the group section. However, to allow an analysis of the specificities of the solo calculation compared to the general characteristics of the model, these (re)insurers should provide two sets of answers to the questions raised in the section IM.C: one set would be the same for all the (re)insurers of the group and should describe the general characteristics of the group internal model and the second set would, if relevant, describe the differences between these generalities and the characteristics of the group internal model when it is used for the solo SCR calculation.
(3)Groups willing to apply for an internal model for the group calculation should answer to questions raised in the sections IM.B, IM.C and in the group section.
(Re)insurers which plan to build, currently building or using internal models in order to get an approval to calculate SAM SCR or only for internal risk management / (Re)insurers which currently building or using internal models in order to get an approval to calculate SAM SCRSolo (re)insurers or (re)insurers which are part of a group but solo internal model is not based on a internal model used for the calculation of the group SCR / IM B / IM B
IM C
Solo (re)insurers which are part of a group and solo internal model is based on/part of a group internal model / IM B / IM B
IM C (two sets of answers, see above, point 2)
IM D
Groups / IM B / IM B
IM C
GIM (Group section)
IM.BQuestions for (re)insurers (both solo entities and groups) which plan to build, are currently building or already use internal models in order to get an approval to calculate SAM SCR or only for internal risk management.
QIM1.Please identify which sections are applicable for you: IM.B/ IM.C/IM.C (two sets of answers)/ IM.D/Group section
QIM2.If you are a solo (re)insurer which is part of a group but the solo internal model is not based/part of the internal model used for the calculation of the group SCR please provide a brief rationale for building a separate internal model.
<QIM2
</QIM2
QIM3.If you are a solo (re)insurer which is part of a group and you intend to calculate the solo SCR from the internal model used to calculate the group SCR, has the internal model been created:
- at your group level or
- at a higher group level
- by another (re)insurer?
QIM4.Are you already using internal models for some individual aspects of your business?
QIM5.If no, are you currently developing an internal model for comprehensive use in managing your business?
QIM6.If you are planning to apply to get your internal model approved in order to calculate the SCR
(a)are you currently working on the implementation of your internal model for SAM purposes?
(b)if no, could you provide the potential date of the beginning of such work?
QIM6b
</QIM6b
(c)could you provide the planned date of submitting the application?
QIM6c
</QIM6c
(d)if you consider that your risk profile deviates from the assumptions underlying the standard formula, please provide the main reasons for this (e.g. deviations in terms of risk exposure, deviations in terms of volatility, non-linear dependency of risks, presence of cycles, incompatibilities of your risks with the SCR modular approach, other – please specify).
QIM6d
</QIM6d
(e)could you please identify risk modules that might lead to inappropriate capital requirements if the standard formula is adopted?
QIM6e
</QIM6e>
QIM7.If the internal model will be used only for internal risk management, under what circumstances would you reconsider this decision (i.e. to use an internal model rather than the standard formula)?
QIM7
</QIM7
IM.CQuestions for insurers or reinsurers which are currently building or already using an internal model for assessing economic capital and for which they plan to apply for approval to use to calculate the SCR under SAM (both solo entities and groups)
C.1Scope of the internal model
IM.C.1Scope of the internal model:
(Re)insurers may model:
- One or more or all risk modules for the whole business;
- One or more risk or all modules for one or more or all major business units;
- One or more risk sub-modules for the whole business, in the same or different risk modules;
- One or more risk sub-modules, in the same or different risk modules, for one or more major business units;
- The adjustment for the loss-absorbing capacity of technical provisions and deferred taxes for the whole business or for one or more major business units;
- The capital requirement for operational risk for the whole business or for one or more major business units.
(re)insurers may use different risk categorizations than those in the standard formula. For example, they may decide to model risks not covered by the standard formula. Internal models do not need to follow a modular structure.
For questions QIM8 to QIM57, the first answer should refer to the 1st set of answer (please, see the second point of the Guidance how to fill in the questionnaire) “set1” (same for all the (re)insurers of the group and should describe the general characteristics of the group internal model) and the second answer to the 2nd set of answer “set2” (In the Excel file, if relevant, describe the differences between the generalities of column D and the characteristics of the group internal model when it is used for the solo SCR calculation).
QIM8.What is the current scope of the internal model? In particular which risks / major business units / entities are included in your internal model?
<QIM8_set1
</QIM8_set1
<QIM8_set2
</QIM8_set2
QIM9.Please compare the structure of your internal model with that of the standard formula. For instance, which risk modules of the standard formula are a) combined, b) divided in your partial internal model.
<QIM9_set1>
</QIM9_set1>
<QIM9_set2>
</QIM9_set2>
For (re)insurers using or intending to use partial internal models (PIM):
QIM10.Please justify the scope of your partial internal model highlighting which criteria you used to define major business units (if relevant) and to choose the scope of the partial internal model (e.g partial internal model as transitory step to full internal model / risk not covered by the standard formula - which ones?/ certain risks covered adequately by standard formula (e.g. a non-life (re)insurer could have a premium and reserve PIM but rely on the standard formula for market risks) – which ones?/ business units not covered adequately by standard formula – which ones?/ acceptable trade off taking into account costs and benefits - for which risks?/ not enough data for some of the risks – which ones?/ an absence of a model which better reflects the risk profile - for which risks?/ other, please specify)
<QIM10_set1>
</QIM10_set1>
<QIM10_set2>
</QIM10_set2>
IM.C.3External models and data
QIM11.If you worked with consultants to develop your internal model, for which specific part of the implementation?
QIM11a - Yes/No in the excel file
<QIM11_set1>
</QIM11_set1>
<QIM11_set2>
</QIM11_set2>
QIM12.If you use external models / data in your internal model, which ones and for which risks / major business units / entities?
QIM12a - Yes/No in the excel file
<QIM12_set1>
</QIM12_set1>
<QIM12_set2>
</QIM12_set2>
QIM13.In which areas were amendments, which adjust the external model / data to your risk profile, necessary?
<QIM13a_set1>
</QIM13a_set1>
<QIM13a_set2>
</QIM13a_set2>
If you identified material and/or quantifiable external model / data limitations and risks arising from the use of external models / data, please describe them briefly.
QIM13b -Did you identify the external model / data limitations and risks arising from the use of external models / data? Yes/No in the excel file
QIM13c - Are those limitations and risks material and/or quantifiable? Yes/No in the excel file
<QIM13d_set1>
</QIM13d_set1>
<QIM13d_set2>
</QIM13d_set2>
IM.C.4Internal model changes
QIM14.If you already have in place a process to develop the policy for internal model changes, how did you distinguish major and minor changes?
QIM14a - Yes/No in the excel file
<QIM14b_set1>
</QIM14b_set1>
<QIM14b_set2>
</QIM14b_set2>
If no, do you have any ideas about criteria which can be applied to distinguish between them?
<QIM14c_set1>
</QIM14c_set1>
<QIM14c_set2>
</QIM14c_set2>
QIM15.If you expect regular major changes to your internal model due to your yearly planning processes, in which areas or with respect to which circumstances.
QIM15a - Yes/No in the excel file
<QIM15b_set1>
</QIM15b_set1>
<QIM15b_set2>
</QIM15b_set2>
IM.C.5Use test
QIM16.How do you plan to demonstrate that the use of your internal model is sufficiently material to result in pressure to improve the quality of the internal model?
<QIM16_set1>
</QIM16_set1>
<QIM16_set2>
</QIM16_set2>
QIM17.Please provide some examples on to what extent is your internal model widely used and plays an important role in the system of governance, risk management and decision making.
QIM17a – Small/Medium/Large degree in the excel file
<QIM17_set1>
</QIM17_set1>
<QIM17_set2>
</QIM17_set2>
How would you demonstrate that persons who effectively run the (re)insurer take into account outcomes of the internal model in building and developing the business strategy?
<QIM17c_set1>
</QIM17c_set1>
<QIM17c_set2>
</QIM17c_set2>
QIM18.If your risk management strategy considers the results produced by your internal model to a small or medium degree, please identify gaps and if to a large degree please provide examples or a justification for such an assessment.
<QIM18b_set1>
</QIM18b_set1>
<QIM18b_set2>
</QIM18b_set2>
QIM19.Are the outputs of the internal model included in regular reporting for the
(a)administrative, management or supervisory body (monthly / quarterly / half-yearly / annually/ not yet), and
(b)other persons who effectively run the (re)insurer (monthly / quarterly / half-yearly / annually/ not yet)?
QIM20.How do you ensure that your administrative, management or supervisory body and the persons who effectively run the (re)insurer understand the internal model and its limitations? How do you plan to improve this understanding?
<QIM20_set1>
</QIM20_set1>
<QIM20_set2>
</QIM20_set2>
QIM20b - Does the documentation include evidence that all levels of management understand the relevant aspects of the internal model? - Yes/No in the excel file
QIM21.Please indicate where you use your internal model, for example in internal project plans; accounting; financial reporting; budgeting; risk management; capital planning; capital allocation; investment policy; economic capital calculations; regulatory capital calculations; MCEV calculations; remuneration; strategic business decisions; product development; pricing; performance analysis; ALM; reinsurance; bonus setting; assessing other customer benefits; target setting; mergers and acquisitions; own risk and solvency assessment; assessment of risk mitigation; portfolio transfer pricing; investment policy; dividend payments; stress tests; market consistent technical provisions; CoC risk margin; assessment of uncertainty in technical provisions; asset allocation; dynamic hedging; other areas – please specify.