IFoA SP7 common mistakes
IFoA SP7 · Understand the traps and how to avoid them.
Writing a generic answer on reserving methods (chain ladder, Bornhuetter-Ferguson, expected loss ratio) without applying the numbers or facts given in the scenario, so the method description could belong to any insurer.
Candidates memorise method mechanics and reproduce them under time pressure rather than reading the specific data pattern (e.g. a development triangle showing a change in case reserving or a diagonal anomaly) and diagnosing what it implies for that insurer.
What to do instead
Before writing, extract the two or three numerical signals in the question (e.g. rising paid-to-incurred ratio, a spike in a single origin year, a change in claims handling) and structure the answer as diagnosis, method choice, and numerical implication. Practise past-style questions by annotating the data first, then writing only points that reference those numbers.
Treating reinsurance as a single product and describing proportional and non-proportional covers generically, without explaining how the specific structure (quota share vs surplus vs excess of loss with a layer and reinstatements) changes the cedant's net result, capital, and reserving.
Reinsurance is often learned as definitions rather than as a mechanism that alters the cedant's loss distribution. Candidates understate how reinstatement costs, aggregate limits, and attachment points affect net volatility and capital relief.
What to do instead
For every reinsurance question, draw the layer structure and compute the cedant's retained loss for at least two scenarios (a small loss below retention and a large loss through the layer). Explicitly state the effect on net technical provisions, capital requirement, and earnings volatility. Build a one-page template linking structure to balance-sheet and P&L effects.
Quoting solvency figures, capital requirements, or regulatory thresholds from memory (e.g. a specific SCR percentage, risk margin method, or Solvency II ratio) without flagging uncertainty or deriving the figure from the data supplied.
Candidates conflate general principles with specific numbers and, under exam pressure, present recalled numbers as facts. This is penalised because the syllabus tests application and judgement, not recall of unverifiable figures.
What to do instead
Answer with the method and the direction of the effect, and where a number is needed, derive it from the question's data and show the working. If a regulatory figure is genuinely required, state the principle (e.g. SCR calibrated to a 1-in-200 one-year VaR) rather than an invented precise number, and note any assumption.
Ignoring the interaction between asset and liability cash flows, and answering ALM questions by listing asset classes or investment types without linking duration, currency, or liquidity mismatches to the insurer's liability profile.
ALM is often revised as investment management rather than as a matched-book problem. Candidates describe equities, bonds, and property in isolation and fail to connect them to claim payment patterns, inflation sensitivity, or capital charges.
What to do instead
Always start from the liability: state its duration, currency, inflation linkage, and liquidity needs, then assess each asset against those criteria. Use a two-column table (liability characteristic vs asset match) and explicitly identify residual mismatch and how it is managed (e.g. swaps, matching adjustment, holding cash for liquidity).
Producing a risk register or ERM answer that lists risks (market, credit, operational, insurance) without assessing likelihood, impact, controls, or ownership, and without linking to the insurer's strategy or risk appetite.
Candidates reproduce taxonomy from memory and believe naming categories earns marks. Examiners reward prioritisation, quantification where possible, and the governance response, which requires judgement the candidate has not practised.
What to do instead
Use a structured format for each risk: description, source, likelihood and impact rating with justification, existing controls, residual risk, and owner or committee. Explicitly connect the top two or three risks to the insurer's stated risk appetite and capital position, and state what would trigger escalation.
Failing to distinguish between data quality issues and modelling choices, and presenting a model output (e.g. a GLM pricing result or a stochastic reserve distribution) as definitive without discussing assumptions, validation, and sensitivity.
Candidates focus on getting the calculation right and treat the model as the answer. The syllabus tests critical evaluation of data and models, so omitting limitations and sensitivity analysis loses substantial marks.
What to do instead
After any calculation, add a short structured critique: key assumptions, data limitations (completeness, credibility, homogeneity), validation performed, and at least two sensitivities showing how the conclusion changes. Practise writing a three-sentence 'limitations and sensitivity' paragraph for every quantitative question.
Writing pricing answers that focus only on the expected loss cost and expense loading, omitting risk margin, capital cost, reinsurance cost, and the required return on capital, so the premium is understated and the answer is commercially naive.
Candidates learn frequency-severity and expense loading as the core of pricing and treat profit and capital charges as an afterthought, missing that a technically correct loss cost does not produce a viable premium.
What to do instead
Build a standard premium build-up template: expected loss cost, expenses, reinsurance cost net of recoveries, cost of capital or risk margin, target profit, and contingency. Apply it to every pricing question and explicitly state the required risk-adjusted return and how the premium achieves it.
Answering professionalism and ethics questions with abstract statements about integrity and objectivity, without identifying the specific threat, the party affected, the applicable actuarial standard or principle, and the concrete action the actuary should take.
Candidates treat ethics as a soft topic and write generic values statements. The syllabus requires applying the actuarial code to a scenario, which demands naming the threat (e.g. pressure from management, conflict of interest, reliance on unverified data) and the professional response.
What to do instead
Use a fixed structure: identify the ethical issue, name the threat category, state the relevant principle or standard, describe the affected parties, then give the action (escalate, document, decline to sign, seek peer review, inform the client). Practise with short scenarios and write the action as an imperative sentence.
Misallocating time and depth across question parts, writing long essays on early parts and leaving later, higher-tariff parts (often the numerical or judgement-heavy sections) with only a sentence or two.
Candidates are drawn to the parts they know best and lose track of marks per minute. Late parts frequently carry the integrative marks that distinguish a pass from a fail.
What to do instead
Before writing, allocate minutes proportional to marks and note the allocation on the answer sheet. If a part is worth 12 marks, plan roughly 12 minutes and four to six developed points. Leave a two-line skeleton for later parts before starting, so every part receives at least its core answer.
Presenting calculations without a clear audit trail: no stated assumptions, no units, no intermediate steps, and no interpretation of the result, so a correct number earns little credit and an arithmetic slip loses everything.
Candidates rush to the final figure and assume the marker will follow the logic. In a judgement-based exam, the marker awards method and interpretation marks, which are lost when working is compressed.
What to do instead
Adopt a fixed layout: list assumptions, show the formula, substitute numbers with units, state the result, then write one sentence interpreting what it means for the insurer. Even when the final number is wrong, the method and interpretation marks are secured.