Every insurer that operates under Solvency II eventually confronts the same triangulation problem: the Solvency Capital Requirement (SCR) says one thing, the internal economic-capital model says another, and the Own Risk and Solvency Assessment (ORSA) is supposed to reconcile the two while adding a forward-looking, undertaking-specific view. When the three numbers are not aligned by construction, they are aligned in the board pack by narrative — which is neither transparent nor defensible.
This article sets out a framework for keeping the three views coherent while preserving their distinct purposes.
Three views, three questions
The starting point is to be honest about what each measure is for.
- SCR answers a statutory question: what capital does the regulator require, calibrated at 99.5% VaR over one year, using either the standard formula in Delegated Regulation 2015/35 or an internal model approved under Articles 112–127.
- Economic capital (EC) answers a management question: what capital does the enterprise need to survive at a chosen confidence and horizon, using the risk metrics, dependence structure and business definitions the enterprise chooses.
- ORSA answers a governance question: given our business strategy over the plan horizon (typically three to five years), do we hold enough capital under plausible scenarios, and is our overall solvency needs assessment materially different from the statutory SCR?
The three are related — SCR calibration influences EC parameterisation, EC informs ORSA scenarios — but conflating them is where the accidents happen.
Where the numbers naturally diverge
Even a fully coherent framework produces different numbers, for legitimate reasons.
Risk measure. SCR uses one-year 99.5% VaR. Many insurers use expected shortfall at a similar or higher confidence for EC, on the grounds that expected shortfall is coherent and better characterises tail losses (see Beyond Historical VaR). Both metrics can be right; they cannot be equal.
Horizon. SCR is one year. Economic capital can be one year but is often multi-year for long-duration life business or for capital-planning contexts. ORSA is explicitly multi-year. The dependence structure at three years is not the same as at one year — long-tail risks correlate more strongly with macro shocks than short-tail risks do.
Contract boundaries. SCR uses the Solvency II contract-boundary definition; management economics often includes the value of future new business or of the going concern, which is excluded from SCR by design.
Diversification. The standard formula uses fixed correlation matrices. Internal models and EC can use bespoke dependence — copulas, factor models, or scenario aggregation. Both are opinions; both should be validated (see Cross-Asset Stress Testing).
Volatility and matching adjustments, transitional measures. These are Solvency II mechanics with no natural analogue in an economic view. An honest reconciliation shows the SCR both with and without these adjustments.
Building the reconciliation
A defensible reconciliation is a bridge, not a table. From SCR to EC, and from EC to ORSA capital need, each step should quantify one specific driver.
SCR (one-year, 99.5% VaR, statutory)
± Risk-measure adjustment (VaR → ES, if applicable)
± Horizon adjustment (one-year → multi-year)
± Contract-boundary adjustment
± Diversification-methodology delta
± Volatility/matching-adjustment delta
= Economic capital (one-year, ES, management view)
± Business-plan projection (new business, dividends, capital actions)
± ORSA stress overlay (severe-but-plausible strategic scenarios)
= ORSA solvency need
Each ± line should have a paper trail: the driver, its magnitude, its rationale and its owner. The document that maps this bridge is more important than any single row of it.
The dependence structure trap
The place where SCR, EC and ORSA most often diverge without anyone noticing is the correlation structure. The standard formula's correlations are calibrated to a specific 99.5% VaR interpretation and are not conditional correlations under stress. An EC model that uses the same numeric correlations in a copula produces materially different tail behaviour depending on the copula family — Gaussian versus t-copula differences at the 99.5%+ tail can be 10–30% of aggregate SCR for a diversified insurer.
If EC uses a t-copula, ORSA stress scenarios have to be coherent with that assumption; running Gaussian-implied scenarios through a t-copula aggregation produces internally inconsistent numbers. A validation function should be able to trace the same dependence assumption through all three views.
Risk appetite: the number that stops the drift
Reconciliation without a risk-appetite framework is an exercise in accounting. Boards want to know not just what the capital is, but what capital position the enterprise chooses to defend. A structured risk-appetite statement translates:
- Regulatory anchor: SCR coverage ratio target (e.g. above 150% under baseline, above 120% under a defined severe scenario).
- Economic anchor: EC coverage ratio target under the internal risk metric.
- Volatility limit: maximum permitted year-on-year swing in SCR coverage.
- Downside limit: coverage ratio at which management actions are triggered — typically dividend suspension, reinsurance activation, capital raising.
When these limits are set consistently with the reconciliation bridge, ORSA scenarios have a clear pass/fail interpretation, and the ORSA report reads as a strategic document rather than a compliance artefact.
Governance implications
The reconciliation is a governance artefact as much as a numerical one. It forces the risk function, the actuarial function, finance and the internal-model owner to agree on definitions before scenarios are run. EIOPA's ORSA guidelines (EIOPA-BoS-14/259) explicitly expect the "overall solvency needs assessment" to consider all material risks, including risks partially or not captured by the SCR calculation — which places climate, reputational, strategic and emerging risks in ORSA even when they sit outside SCR.
Board reporting should present all three views side by side, with the bridge, and with a plain-English commentary on the drivers of change quarter on quarter. A single-number solvency slide invites the wrong questions.
Validation and independent challenge
Validators should attack the reconciliation from both ends. From the SCR side: are internal-model parameters that inform EC re-anchored consistently to statutory calibrations at least annually? From the EC side: does the internal model produce SCR-equivalent numbers when its assumptions are dialled back to the statutory ones? The gap between those two numbers, controlling for known drivers, measures the model risk in the reconciliation itself. See The Model Validation Lifecycle for the broader framework.
Limitations
No reconciliation exercise removes model uncertainty; it makes uncertainty explicit. Insurers that treat EC as a superior "true" number to SCR miss the point — EC is a management opinion, and its confidence intervals are wider than most internal documents acknowledge. Insurers that treat SCR as the sole capital number miss the strategic view that only ORSA can provide.
Conclusion
The three views are complementary. Kept coherent by an explicit reconciliation bridge and a risk-appetite framework, they give the board, the regulator and the internal-model function distinct answers to distinct questions. Kept implicit, they invite the number the audience wants to hear rather than the one the risk profile deserves.
Related reading
References and further reading
- European Parliament and Council, Directive 2009/138/EC on the taking-up and pursuit of the business of Insurance and Reinsurance (Solvency II).
- Commission Delegated Regulation (EU) 2015/35 supplementing the Solvency II Directive.
- European Insurance and Occupational Pensions Authority, Guidelines on own risk and solvency assessment (EIOPA-BoS-14/259).
- International Association of Insurance Supervisors, Insurance Core Principles.
About the author
Part of an ongoing series on insurance capital and risk modelling — more about the author and Quantica Risk Modelling.