Skip to content
←Back to glossaryActuarial science

Risk loading

Amount added to pure premium to remunerate tied-up capital and absorb volatility, without which expected profit is nil and long-run ruin is certain.

Definition

Filip Lundberg set out in 1903, in his Uppsala thesis, the collective risk model from which ruin theory draws a classical result: a premium equal to expected loss leads to ruin with probability one over an infinite horizon. Surplus then follows a driftless walk, and a driftless walk hits zero. The risk loading is the drift you add. It pays for the capital the shareholder ties up to absorb deviations; its correct size therefore depends on the capital the risk consumes, and not on its expectation. Three calculation principles coexist and give different answers. A loading proportional to expectation is simple and blind to volatility, so it underprices dispersed risks. A loading proportional to standard deviation recognizes volatility but not dependence with the rest of the book. A cost of allocated capital loading applies a return rate to the capital the risk consumes after diversification; it alone is consistent with balance sheet management, and internal models have converged on it. Expense loadings, for their part, cover spending rather than risk, and are not to be confused with the risk loading.

Its 8 cross-references, woven across the glossary's 18 categories
Risk loading: Actuarial science glossary illustration, AlgoPolis
Example

Treaty covering a risk with a 10M EUR expectation and a 6M EUR standard deviation, 2026 year. A 5% loading on expectation returns 500K EUR. A 10% loading on standard deviation returns 600K EUR. A cost of capital loading, with 18M EUR of allocated capital after diversification and a 9% internal return rate, returns 1.62M EUR. The three principles applied to the same risk spread the gross premium by more than nine combined ratio points.

Read next
Related terms
Related articles
Stay informed

Follow the research

One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.

No spam, one-click unsubscribe, no data ever sold.