Authors: Jasper Hoogenstraaten, Servaas Houben
link: http://theeuropeanactuary.org/downloads/TEA%2021-nov2019.pdf (pages 8-11)
pdf: TEA 21-nov2019_CoC
Publisher, publication date: The European Actuary, 2019-11
In the previous TEA article on the IFRS 17 risk adjustment as part of the technical provisions we have argued that the differences between Solvency II and IFRS 17 will provide insurance companies with room for own interpretation under IFRS 17 as the risk adjustment should be a reflection of the company’s risk profile and their own risk appetite. The Cost of Capital (CoC) method and Confidence Interval (CI) method are two common techniques for determining the risk adjustment as a buffer for non-financial risks. In this article we will look at a practical example to assess the impact on the choice of methodology for risk adjustment. Although IFRS 17.119 does prescribe disclosure of the CI level corresponding to the risk adjustment value, we will argue that using the CoC method to actually determine the risk adjustment will still provide additional insights and might better align the current and future balance between risk and capital of the specific insurance company involved.
IFRS 17 opportunities
As mentioned in the previous article the Solvency II risk margin is based on the principle of transferring liabilities to a third party while the IFRS 17 risk adjustment is based on the principle of a going concern.
|Solvency II Risk Margin||IFRS 17 Risk Adjustment|
|Valuation perspective||Transfer to third party||Going concern own entity|
|Scope||All relevant SCR risks including operational risk and non-hedgeable financial risk||Contract specific non-financial risk only|
|Valuation method||Cost of capital||Own estimation technique|
|Stress level||99.5% following SCR||Dependent on company’s own degree of risk aversion|
|CoC rate||6%||Not predefined, can be company specific or other method may apply|
|Shock type||Unfavourable outcomes||Assess risk aversion to favourable and unfavourable outcomes|
Although IFRS 17 does not prescribe a methodology to determine the risk adjustment and allows for a more company specific interpretation, IFRS 17. B91 does lists the following characteristics a risk adjustment metric would require:
- Low frequency high severity risks should reflect in a higher risk adjustment compared to high frequency low severity risks. E.g. an uncommon miss-selling event would result in a higher risk adjustment than a frequent IT failure;
- Risks with longer durations would reflect in a higher risk adjustment. E.g. lifelong annuities would require a higher risk adjustment compared to temporal annuities;
- The wider the distribution of the risk, the higher the risk adjustment. The more uncertainty around risks, would require a higher buffer;
- The less information is known about the current risk and future trends the more risk adjustment is needed. E.g. in countries with little mortality (trend) data the risk adjustment would be higher than countries with more data;
- When emerging experience reduces uncertainty, the risk adjustment decreases. During the term of a temporal annuity portfolio, it will become clearer what the actual vs expected cash flows will be and hence the risk adjustment will decrease.
As both the CI and CoC method take into account the requirements mentioned above, they are both suitable techniques to determine a risk adjustment.
Suppose a lifelong annuity product from age 65 with annual payments of 1 and a flat discount rate of 3%. For this example the following capital costs apply:
|Method||Cost of Capital|
|Cost equity financing||10%|
|Cost debt financing||6%|
|Allocation equity financing||40%|
|Allocation debt financing||60%|
|Weighted average cost of capital||7,6%|
Table 1: cost of capital assumptions
One of the main risks for such a product potentially resulting in the cash flows to deviate from the best estimate cash flows is mortality.
We assess the impact of a mortality level down stress with the following shocks and distribution assumptions:
|Longevity standard dev||0,1|
Table 2: distribution parameters and shock assumptions
The CI method is based on a projection of the technical provision at the chosen confidence level (i.c. 70%) and implicitly only applies a stress to the BEL at time zero will result in the following capital requirement:
|Time||ax (BEL)||ax (down)||Capital requirement|
Table 3: CI capital requirement
The CoC method instead applies a stress for each time during the term of the contract. Figure 1 below compares the best estimate reserve to the stressed reserve, the difference between the two being the capital requirement.
Figure 1: CoC projection of reserves and capital requirement over time
The figure above displays a decreasing capital requirement for mortality risk during the lifetime of the policy. These future capital requirements are finally transformed to a cost of capital for each future period by determining the net present value of all future capital requirements multiplied by the WACC%. In this example this would result in a total of 2,138.
Not surprisingly, in the case of an annuity – a product with a long duration – the CoC calculation, explicitly taking into account a stress at future periods, will result in a higher capital requirement compared to the CI method.
Suppose however that the insurance companies long term capital strategy is to increase its financing by means of equity from the current 40% to a level of 60% over a period of 20 years (an increase of 1%-point per year). This has an impact on the WACC since equity financing is considered have a higher cost of capital following table 1. While the CI level will provide the same risk adjustment number (as only the CI level at time 0 is taken into account) the CoC method will result in a different risk adjustment instead (i.c. an increase to 2,274) reflecting the change in capital structure during the projection period. Therefore the CoC method seems to better reflect the changing environment of the insurance company over time.
IFRS 17 allows insurance companies to use their own risk appetite and CoC interpretation while determining the risk adjustment. This will not only allow them to let the risk adjustment be more in line with their own view on risks but will also allow them to potentially adjust their capital funding to optimize the balance between capital and risk.
The straight-forward example discussed in this article shows that with limited effort and analysis, a substantial insight can be derived in both risk assessment and capital funding. Extending the analysis with assumptions and views on different risks and diversification benefits, will further enhance the understanding of the interaction between risks and capital.
Although IFRS 17.119 prescribes companies to disclose their risk adjustment on a CI level, the CoC approach is still a valuable alternative as it takes into account the duration of the liabilities and potential changes in capital funding and can therefore be a longer term counterweight to more short term measures.
- Damodaran information on weighted average cost of capital,
- http://www.stern.nyu.edu/~adamodar/pc/datasets/wacc.xls (US, similar setup for other geographies)
- EIOPA’s second set of advice, https://eiopa.europa.eu/Publications/Consultations/EIOPA-18-075-EIOPA_Second_set_of_Advice_on_SII_DR_Review.pdf
- EIOPA risk free interest rate curves, https://eiopa.europa.eu/regulation-supervision/insurance/solvency-ii-technical-information/risk-free-interest-rate-term-structures
- Final CEIOP’s advice for Level 2 implementing measures on Solvency II: technical provisions – article 86 (d) calculation of the risk margin, https://eiopa.europa.eu/CEIOPS-Archive/Documents/Advices/CEIOPS-L2-Final-Advice-on-TP-Risk-Margin.pdf
- Solvency II Directive, https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:02009L0138-20140523&from=EN
 Confidence interval i.e. compared to the mean of 50%