IAI SP8 common mistakes
IAI SP8 · Understand the traps and how to avoid them.
Treating the chain-ladder development factors as an estimate of the ultimate loss ratio and quoting them as if they were a priced expectation, rather than applying them to reported claims and then adding a separately justified IBNR/initial expected loss ratio load for immature years.
Chain-ladder is mechanical and easy to compute, so candidates often stop at the projected ultimate and never articulate the assumptions (stable development pattern, no material change in case reserving or settlement speed) or the need for a Bornhuetter-Ferguson style blend on the most recent accident years where the data are thin.
What to do instead
For every reserve question, write the three assumptions explicitly, show the development triangle pattern, and state whether you are using chain-ladder, Bornhuetter-Ferguson or a frequency-severity approach and why. For the latest year, present the initial expected loss ratio and the resulting IBNR as a range, and reconcile the total reserve to the prior year with a movement analysis (paid, incurred, prior-year development, discounting).
Calculating a gross premium by adding a flat expense loading and a flat profit margin to the expected loss cost, then presenting that as the technical price without testing the sensitivity of the loss ratio to a plausible range of loss assumptions.
Candidates default to the simple expected-value formula and forget that the expense ratio is a percentage of premium, so the loading must be solved algebraically (pure premium divided by 1 minus expense and profit ratios), and they omit the risk margin and the cost of capital that a risk-adjusted price requires.
What to do instead
Derive the price as pure premium / (1 - variable expense ratio - fixed expense per policy / premium - profit and contingency margin), then show the loss ratio and combined ratio implied by a 1-in-200 and a 1-in-20 loss scenario. State the risk load explicitly and explain how it relates to the cost of holding capital.
Describing reinsurance as simply 'transferring risk' and quoting the ceded premium as the cost of the cover, without computing the expected recovery, the retained loss after the retention, or the effect on the net combined ratio and capital requirement.
Reinsurance answers are often written qualitatively because the arithmetic is fiddly, so candidates avoid the numerical comparison and never show that proportional and non-proportional structures have very different effects on volatility versus expected cost.
What to do instead
For each structure, build a small table: gross loss, retention, ceded loss, reinsurance premium, net loss, and net result under a normal year and a catastrophe year. Compute the expected ceded loss and compare it to the ceded premium to show the loading, and comment on how the structure reduces the tail and therefore the solvency capital requirement.
Presenting a single deterministic solvency ratio and concluding the insurer is adequately capitalised, without a stress or scenario test and without explaining how the capital requirement was derived.
Candidates memorise the structure of a risk-based capital regime but do not practise applying shocks, so they report the base ratio and stop, missing that the exam wants evidence of understanding of the calibration, the aggregation (correlation) assumption and the management actions available.
What to do instead
Always present base ratio, then at least two stresses (a large loss event and a reserve deterioration or asset shock), and state the post-stress ratio against the regulatory control level. Explain the aggregation assumption, the treatment of diversification, and which management actions (reinsurance, capital raise, asset de-risking) would restore the ratio and how quickly.
Confusing the roles of assets and liabilities in an ALM answer, for example recommending a long-duration bond portfolio for a short-tail motor book, or ignoring that reserves are nominal and undiscounted while assets are marked to market.
ALM is taught as a separate topic and candidates do not link it back to the liability duration implied by the reserving and claims pattern, so the duration mismatch and the interest-rate sensitivity of the surplus are never quantified.
What to do instead
Start from the liability cash-flow profile implied by the claims settlement pattern, state the liability duration, then match asset duration and cash flows to it. Quantify the surplus sensitivity to a 100 basis point shift in yields, and note the accounting mismatch between discounted assets and undiscounted reserves where relevant.
Writing a professionalism answer that names the code of conduct and then asserts the actuary 'should act ethically', without identifying the specific threat, the affected parties, the safeguards available and the escalation route.
Candidates treat ethics as a soft topic and reproduce generic statements, but the marks are in the structured analysis: self-interest or intimidation threat, who is disadvantaged, what the actuary should document, and to whom the concern should be reported if internal escalation fails.
What to do instead
Use a fixed structure: facts, the threat category, the principles engaged (integrity, competence, objectivity, confidentiality), the affected stakeholders, the options with consequences, the recommended action, the documentation, and the escalation path including the professional body if the concern is not resolved.
Listing risk categories (underwriting, credit, market, operational, liquidity) and stopping there, without mapping each to a measurable exposure, a quantification method and a mitigation, and without distinguishing risk identification from risk measurement.
The domain is broad and candidates produce a taxonomy as if naming categories were the answer, whereas the exam rewards showing how each risk is measured (frequency-severity, scenario, stress) and how it feeds the capital model.
What to do instead
For each material risk, write three lines: source and driver, measurement approach with the data required, and mitigation or transfer. Then state which risks are captured in the internal model, which are covered by the standard formula, and where the model may understate the exposure (for example concentration or aggregation risk).
Reporting the technical result and the investment return separately and then adding them, without reconciling the total result to the movement in shareholders' equity or explaining the effect of the change in reserves and the change in unrealised gains.
Candidates know the components of the income statement but not the articulation between the underwriting result, the investment return on reserves and capital, and the balance sheet movement, so the performance analysis does not close.
What to do instead
Build a bridge: opening equity, underwriting result, investment income allocated between policyholder and shareholder funds, other items, tax and dividends, closing equity. Show the combined ratio, the investment return on the float, and the return on capital, and explain which driver moved the result year on year.
Describing the general insurance market generically ('the market is competitive and cyclical') without connecting the underwriting cycle to capacity, reserving adequacy and pricing behaviour, or to the specific line of business in the question.
Market knowledge is often memorised as a list of participants and distribution channels, so candidates do not use the cycle as an analytical tool, and they fail to link softening prices to the reserve releases that flatter reported results.
What to do instead
Anchor the answer in the specific line: who the capacity providers are, how distribution works, what drives claims inflation in that line, and where the cycle currently sits. Then link it to the pricing decision (whether the technical price is being met) and to reserving (whether prior-year releases are masking deterioration).
Producing a long narrative answer with no numbers, no stated assumptions and no conclusion, so the marker cannot award the application and judgement marks even where the theory is correct.
Under time pressure candidates write everything they know about the topic rather than answering the question asked, and they omit the explicit assumptions and the recommendation that distinguish a pass from a distinction.
What to do instead
Before writing, allocate marks to a plan: define, calculate with stated assumptions, interpret, recommend. Keep the calculation visible with units and rounding stated, and finish with a short conclusion that directly answers the question and names the key uncertainty. Practise under timed conditions so the plan becomes automatic.