Product design, pricing and underwriting for general insurance
This domain tests pricing and underwriting technique in a commercial and regulatory frame: how loadings are derived, how they are defended, and where judgement replaces data.
Core technique
- Decompose expected loss into frequency and severity, or into loss ratios by homogeneous risk group; choose exposure bases (payroll, turnover, vehicle-years) that respond to risk.
- GLMs (Tweedie for pure premium, Poisson for frequency, Gamma for severity) allow many rating factors without overfitting. Credibility blends a thin book with a benchmark, using Z = n/(n+k).
- Technical price reflects expected loss, expenses, commission, capital cost and profit; market price reflects the cycle and demand. Present both and explain the gap.
Underwriting and terms
- Excess, aggregate limits, sub-limits, exclusions and conditions each transfer risk. Under the Insurance Act 2015 a warranty operates as a suspensive condition, not a right to avoid from inception.
- Price optimisation using demand elasticity must survive the fair-value test; the FCA's general insurance pricing rules ban the loyalty penalty on home and motor renewals (from 1 January 2022).
Worked example
A book shows a five-year burning cost of 40 per policy; a benchmark for comparable risks is 60, and credibility Z = 0.3. Blended loss cost = 0.3 x 60 + 0.7 x 40 = 46. Expenses and commission take 25% of premium, so the risk-loaded technical premium is 46 / 0.75 = 61.3 before any profit or capital charge. The exam marks are in justifying Z (volume and stability), the direction of the benchmark adjustment, and placing the expense loading on premium rather than on loss cost.
Traps
Dividing by (1 - expense ratio) when expenses are quoted as a percentage of loss cost; applying credibility without explaining k; ignoring that a thin questionnaire pushes risk into the unexpired risk provision.