Why One Stress Can Produce Different Capital Outcomes
An enterprise stress scenario describes adverse conditions and their effects on an organization. It does not, by itself, determine a single capital result. An internal capital assessment, a regulatory framework, and a rating agency assessment may begin with a similar economic narrative yet reach different results because they may measure different exposures, apply different calibrations, aggregate risks differently, or recognize different resources.
The useful comparison is therefore not simply “which number is larger?” It is “what does each number measure, and which assumptions produced it?” A capital requirement, a post-stress resource amount, a coverage ratio, and a management target are distinct quantities. They should not be treated as interchangeable just because each is expressed in dollars or as a ratio.
The internal view is designed to support the organization’s own risk and capital decisions, using its chosen model scope, assumptions, and risk appetite. A regulatory view applies the definitions and requirements of the relevant jurisdiction. A rating agency view applies the rating agency’s methodology and analytical scope. The details vary across organizations, jurisdictions, and agencies: there is no single universal regulatory formula or rating agency capital formula.
A comparison should hold the economic scenario as constant as practicable, then identify where each view transforms it differently. Earlier material on enterprise scenario architecture, scenario calibration, aggregation, and capital-resource bridges provides the tools to describe those transformations. The focus here is how to interpret the resulting differences across views.
A Reconciliation Framework
For each view, document at least four items: scope, calibration, aggregation, and eligible resources. A fifth item, the reported outcome measure, is necessary to interpret the result correctly.
| Dimension | Question to ask | Potential effect on the result |
|---|---|---|
| Scope | Which entities, exposures, risk categories, and time horizons are included? | A broader or narrower scope changes the risks being measured. |
| Calibration | How severe is the shock, and what horizon or confidence standard is used? | A more severe or differently timed stress can change losses and requirements. |
| Aggregation | Are risks added, diversified, or combined using a specified model? | Recognized diversification can lower an aggregate relative to a simple sum. |
| Eligible resources | Which assets, own funds, recoverables, and other resources count? | Limits, deductions, and haircuts affect the resources available to absorb loss. |
| Outcome measure | Is the reported number a requirement, resources, headroom, or a ratio? | Different measures cannot be ranked without first reconciling their meanings. |
A common bridge expresses post-stress resources as opening eligible resources less stress losses, with any separately recognized resource changes shown explicitly. If \(R_0\) is opening eligible resources, \(L\) is the stress loss, and \(M\) is the net change in recognized resources under the stress, then:
Here \(M\) may be positive or negative. It could represent a recognized mitigation or a further deduction, but it must be defined consistently with the view being analyzed. If a capital requirement \(K\) is also provided, headroom is \(R_{\text{stress}}-K\). A positive value means resources exceed that requirement under the stated definitions; a negative value means a shortfall against it. This arithmetic does not imply that the different frameworks use the same requirement or define resources in the same way.
Worked Example: Scope and Diversification Change the Aggregate
Worked Example: Scope and Diversification Change the Aggregate
A hypothetical insurer applies one enterprise stress and estimates two internal risk charges of $60 million and $80 million. Its internal aggregation uses a variance-covariance method with an assumed correlation of \(0.25\). A simplified regulatory comparison adds the same two charges without diversification. A hypothetical rating agency view includes the same stressed risks plus $12 million and $8 million of additional in-scope charges, respectively, and adds all charges. Calculate and compare the three totals. Amounts are in millions of dollars.
Method. Apply the stated aggregation rule separately in each view. The variance-covariance formula for two charges \(C_1\) and \(C_2\), with assumed correlation \(\rho\), is \(\sqrt{C_1^2+C_2^2+2\rho C_1C_2}\). It is appropriate here only because the problem explicitly specifies that method and its correlation assumption.
Internal view. Substitute \(C_1=60\), \(C_2=80\), and \(\rho=0.25\):
The internal aggregate is approximately $111.3553 million. The radicand is $12,400, not $10,000: the cross-term is \(2(0.25)(60)(80)=2{,}400\).
Regulatory comparison. The stated rule is direct addition without diversification:
The simplified regulatory total is $140 million, which is $28.6447 million greater than the internal aggregate: \(140-111.3553=28.6447\).
Rating agency view. First include the additional in-scope charges: \(60+12=72\) and \(80+8=88\). Then apply the stated additive rule:
The rating agency total is $160 million, $20 million above the regulatory comparison and $48.6447 million above the internal aggregate. The differences follow from the stated aggregation and scope rules; they do not show that the economic stress itself changed.
Check. The internal total lies between the largest individual charge ($80 million) and the undiscounted sum ($140 million), as expected for this positive-correlation variance-covariance calculation. The rating total equals the two original charges plus the two additional charges: \(60+80+12+8=160\). A common mistake is to assume every framework must use the internal correlation; the problem explicitly gives different rules for the comparison views.
Why Aggregation Produces a Different Number
The two-risk variance-covariance expression follows from the variance of a sum. If centered risk variables \(Z_1\) and \(Z_2\) have standard deviations \(C_1\) and \(C_2\), and correlation \(\rho\), then:
Taking the square root gives the standard deviation of the sum and, when a framework defines its aggregate that way, the corresponding aggregate charge. This derivation explains the cross-term: it represents the dependence between the two risks. It does not establish that the formula is appropriate for every capital view. In particular, it should not be substituted for a prescribed rule, or used to infer a full loss distribution from only two charges and a correlation.
The example’s assumed correlation is a modeling choice, not a universal fact about the risks. Also, correlation-based aggregation and direct addition are not merely alternative ways to report an identical calculation: they encode different treatments of diversification. A comparison should name the rule rather than describe one total as inherently correct in all contexts.
Worked Example: Reconcile Resources, Requirements, and Headroom
Worked Example: Reconcile Resources, Requirements, and Headroom
A hypothetical group evaluates the same broad stress in three frameworks. Its internal view starts with $300 million of eligible resources, applies a $125 million stress loss, and compares the result with an internal capital target of $150 million. A regulatory view starts with $280 million of eligible own funds after its framework’s opening eligibility adjustments, applies a $140 million stress loss, and compares the result with a $130 million requirement. A rating agency view starts with $260 million of eligible resources under its own definitions, applies a $150 million stress loss, and compares the result with a $120 million capital benchmark. Calculate each post-stress resource amount, headroom, and, for the regulatory and rating views, resources-to-requirement ratio.
Method. For each view, subtract its stated stress loss from its own opening eligible resources. Then subtract that view’s stated requirement from post-stress resources to find headroom. Compute a ratio only where a requirement is supplied and express it as post-stress resources divided by that requirement.
Internal view. Post-stress resources are \(300-125=175\) million. Headroom to the internal target is \(175-150=25\) million.
Regulatory view. Post-stress eligible own funds are \(280-140=140\) million. Headroom is \(140-130=10\) million. The resources-to-requirement ratio is \(140/130=1.076923\ldots\), or \(107.69\%\) to two decimal places.
Rating agency view. Post-stress eligible resources are \(260-150=110\) million. Headroom is \(110-120=-10\) million, a $10 million shortfall. The ratio is \(110/120=0.916666\ldots\), or \(91.67\%\) to two decimal places.
Check. The regulatory ratio exceeds 100% because $140 million exceeds $130 million; the rating ratio is below 100% because $110 million is less than $120 million. These outputs are not directly interchangeable: each uses different opening eligible resources, losses, and thresholds. A common mistake is to compare $150 million, $130 million, and $120 million as if all three were resource amounts. They are requirements or benchmarks, not post-stress resources.
Calibration and Resource Recognition Need Separate Bridges
Even when a scenario has the same name, calibration may differ. A “severe market decline,” for example, does not specify the size, duration, or joint movement of the shocks. One view may apply a one-year stress, while another may use a prescribed stress definition or a different horizon. Before attributing a capital difference to diversification, confirm that the loss inputs actually represent comparable shocks.
Resource eligibility can also change the result without changing gross economic loss. Frameworks may differ in whether, and to what extent, they recognize a resource, a recoverable, or another loss-absorbing item. A disciplined bridge records the gross loss, the recognized offset or deduction, and the resulting net loss separately. Do not silently net an item in one view while leaving it gross in another.
Worked Example: Different Recognition of a Stress Offset
A hypothetical stress produces a $100 million gross loss. The opening eligible resources are $250 million under an internal view, $230 million under a regulatory view, and $225 million under a rating agency view. For this illustration, an offset associated with the stress is recognized at $20 million, $15 million, and $10 million, respectively. Each view subtracts its recognized net loss from opening eligible resources. The respective capital thresholds are $160 million, $140 million, and $145 million. Calculate each net loss, post-stress resource amount, and headroom. Amounts are in millions.
Method. For each view, calculate net loss as gross loss less the view’s recognized offset. Then calculate post-stress resources and headroom. The differing offset amounts are hypothetical assumptions, not statements about the rules of any particular framework.
Internal view. Net loss is \(100-20=80\). Post-stress resources are \(250-80=170\). Headroom is \(170-160=10\).
Regulatory view. Net loss is \(100-15=85\). Post-stress resources are \(230-85=145\). Headroom is \(145-140=5\).
Rating agency view. Net loss is \(100-10=90\). Post-stress resources are \(225-90=135\). Headroom is \(135-145=-10\), a $10 million shortfall.
Check. Each post-stress amount equals opening resources less the stated net loss: \(250-80=170\), \(230-85=145\), and \(225-90=135\). The respective headroom calculations also reconcile to the stated thresholds. The gross loss is identical in all three views; differences in recognized offsets, opening resources, and thresholds explain the different outcomes. A common mistake is to report the $100 million gross loss as the net loss in every view despite the explicitly different offset recognition.
How to Explain a Difference Clearly
A useful explanation separates the numerical reconciliation from the judgment about what it means. First state which scenario and loss measure are being compared. Then identify whether a difference arises from scope, calibration, aggregation, resources, or the outcome definition. If several dimensions differ, quantify them in a bridge where the available information permits; otherwise, state which assumptions prevent a clean attribution.
Confirm the stress narrative, time horizon, and principal shock assumptions being compared.
List included entities, risk categories, exposures, and any additional or excluded items.
Show standalone amounts and the stated treatment of diversification, dependencies, and offsets.
Identify opening resources, eligibility adjustments, stress effects, and recognized mitigations.
Report requirements, post-stress resources, headroom, and ratios as distinct measures, then explain their decision use.
This approach also guards against over-interpreting a ratio. A ratio depends on both its numerator and denominator. A lower ratio can result from lower eligible resources, a higher requirement, or both. Its interpretation is framework-specific, and a ratio from one view cannot automatically be treated as a rating outcome or regulatory solvency measure in another.
Common Comparison Traps
- Assuming the scenario label means identical calibration. Verify the shock size, horizon, and included drivers rather than relying on a shared name.
- Applying one view’s diversification to another. Use each framework’s stated aggregation method. Do not add a correlation adjustment unless the method calls for it.
- Confusing losses, resources, and requirements. A stress loss reduces resources; it is not itself a capital requirement. Headroom is a difference between resources and a stated threshold.
- Netting offsets inconsistently. State whether a mitigation is recognized, and at what amount, in each view.
- Calling a difference a model error without a bridge. A higher or lower number may reflect a deliberate scope or eligibility difference; identify the cause before drawing a conclusion.
Check Your Understanding
For each problem, apply only the assumptions stated and check that the final measure matches what the question asks.
- A hypothetical internal model aggregates charges of $30 million and $40 million using the variance-covariance method with correlation \(0.50\). Calculate the aggregate to four decimal places. (Answer: \(\sqrt{30^2+40^2+2(0.50)(30)(40)}=\sqrt{3{,}700}\approx 60.8276\) million dollars.)
- A hypothetical capital view has $190 million of opening eligible resources, a $55 million stress loss, and a $125 million requirement. Calculate post-stress resources and headroom. (Answer: \(190-55=135\) million post-stress resources; \(135-125=10\) million headroom.)
- A regulatory comparison has $204 million of eligible own funds against a $170 million requirement. Under a stated stress, own funds fall by $24 million and the requirement rises by $10 million. Calculate the post-stress ratio and state whether it is above, equal to, or below 100%. (Answer: \((204-24)/(170+10)=180/180=1.0000\), or \(100.00\%\), equal to 100%.)
- Two risks have charges of $45 million and $35 million. An internal view uses the variance-covariance method with correlation \(0\), while a simplified comparison view adds the charges without diversification. Calculate both aggregates to four decimal places and their difference. (Answer: Internal \(=\sqrt{45^2+35^2}=\sqrt{3{,}250}\approx57.0088\) million; additive \(=45+35=80.0000\) million; difference \(=22.9912\) million.)
- A hypothetical view starts with $300 million of eligible resources, deducts $18 million as an eligibility adjustment, incurs a $72 million stress loss, and recognizes an $8 million stress offset. Its requirement is $200 million. Calculate post-stress resources and headroom. (Answer: \(300-18-72+8=218\) million post-stress resources; \(218-200=18\) million headroom.)