Mildenhall CH12-15 Flashcards
(28 cards)
Two approaches to risk-adjust the target return and fix the amount of capital
- Risk-adjusted return on capital (RAROC) methods that say return varies with risk
- Return on risk-adjusted capital (RORAC) methods that say return is constant when capital reflects risk
Describe RAROC (risk-adjusted return on capital)
When a single constant CoC is combined with a capital allocation that normalizes for risk.
We call this the allocated CCoC pricing method.
What’s the problem with constant CoC?
Since insurers are funded by different types of capital with varying cost characteristics, each unit will consume a unit mix of capital. So even all units have a constant CoC within a layer, the weighted cost of capital across layers will differ since each unit has a different mix of capital.
How to determine the constant CoC to be used
We rely on the estimate of the insurer’s weighted average cost of capital (between debt, reinsurance, and equity)
We can quantify debt and reinsurance costs, so leaving the equity cost of capital as the most unknown input.
3 desirable properties of an allocation.
- It should work at any level of granularity, even down to the policy
- It should be decomposable, meaning that the allocation to a sum of random variable equals the sum of their allocations
- it should be computed using a single, consistent formula
What is endogenous allocation
When the same risk measure is used to determine the total and allocate it.
What is exogenous allocation
When the same risk measure is not used to determine the total risk measure and allocate it. When the amount is stipulated by an external risk measure and allocated using a auxiliary measure.
Occurs when a regulator or rating agency determines the amount of capital, but an insurer determines its allocation.
What allocations is the solution to optimization problem
Equal risk VaR.
Each unit, stand-alone, has an expected loss beyond its allocated capital. The objective is to allocate capital to minimize the sum of these expected excesses.
Describe marginal business unit or Merton-Perold method
it attributes to each unit the reduction in capital from dropping it from the portfolio.
This method is not additive: generally the sum of allocation is less than total capital.
Describe marginal business Euler gradient allocation
It’s based on the marginal change in capital given a marginal change in the amount of unit i written.
This is endogenous allocation.
Why is the Euler gradient allocation is only one suitable for performance measurement
The growing lines with higher return to marginal capital always improve the average return.
This assumes that the average return is sufficient for deciding whether a portfolio’s pricing has improved, it does not allow for the possibility that a change in portfolio composition causes a change in the cost of capital.
What is no-undercut property
an allocation where no unit is allocated more than its stand-alone costs
When is a game called atmoic
when each unit is in or out
When is a game called fractional or fuzzy
when units can form tractional coalitions (quota share)
Desirable properties of Shapley allocation
- It’s an allocation, so it’s additive. In game theory, this is called the efficiency property
- It’s symmetric. if two units i and j increases the cost of every S that contains neither i nor j by the same amount, then cj = cj
- It’s decomposable. It’s called linear in game theory
- It’s homogeneous if c is
- if c is subadditive then Shapley value satisfies the no-undercut property
- It allocates no capital to a constant risk, called the null player or riskless allocation property.
Shapley value is the only allocation that is additive, symmetric, linear and allocates no capital to constant risks
2 drawbacks of Shapley allocation
- To allocate to n units involves computing 2^n marginal impacts, which quickly becomes impractical.
2 If a unit is subdivided further into two new units then allocations assigned to the other units typically change.
Characteristics of Aumann-Shapley allocation
it satisfies linearity, decomposable, and either continuity or non-negativity when a is non decreasing
When is Exogenous allocation helpful
When a is determined by a risk measure that is difficult to allocate.
Two ways to derive premiums using capital allocation methods
- Allocated CCoC pricing (allocate assets or capital to each unit and then use those to calculate premium based on the CCoC formulas)
- Direct Allocation of PCPs (allocate total premium to each unit directly using the allocation methods)
Considerations when choosing an allocation method
- All allocation must be considered fair. It means the stakeholders consider the allocation reflective of reality and not overly influenced by factors not specific to the problem the allocation is supposed to solve.
- The allocation is not overly tail-focused. Aumann-Shapley allocation can produce an overly tail-focused answer. The body of the risk distribution also matters.
- Professional actuarial standards should be followed
- Recognizing the scope of the larger task within which allocation is a part
- The value of simplicity, transparency, fairness, objectivity and precedent.
- Sensitivity to stakeholder concerns.
- Communication and persuasion.
Describe natural allocation of coherent risk measure
It’s an explicit allocation formular that applies to coherent risk measures.
The allocation is natural because it entails no additional choices (no new probability threshold for example).
It follows the finance philosophy to adjust the probabilities and act risk neutral.
When is marginal allocation method useful?
It’s useful for performance management, meaning they give the right grow or shrink signal for portfolio optimization in a constant CoC world.
Why co-measure allocation is useful
It’s intuitive, practical, and easy to apply.
They can be interpreted as risk-adjusted probabilities, and this interpretation provides a philosophical alignment with finance. and they are closely related to the natural allocation.
Requirements for the risk measure for linear natural allocation
The risk measure must be law invariant.
The second constraint restricts attention to risk measures where it does hold.
This can be achieved by assuming the risk measure is comonotonic additive, so meaning the risk measure is spectral risk measure.