IFoA SP8 common mistakes
IFoA SP8 · Understand the traps and how to avoid them.
Answering a calculation question with a single number and no stated assumptions, formula, or intermediate steps, so method marks cannot be awarded even when the final figure is right.
Candidates treat SP8 numerical questions like a spreadsheet exercise and rush to the final figure, forgetting that markers award marks for the approach (formula, parameter choices, intermediate results) and that a bare number cannot be credited if a step is wrong.
What to do instead
Adopt a fixed layout for every calculation: state the formula or model, list each input and its source, show one line per step, box the final answer with units and a one-line interpretation. Practise under timed conditions until this layout is automatic.
Writing generic textbook lists (for example 'consider reinsurance, expenses, investment return') without tying each point to the specific scenario, class of business, or data given in the question.
Under time pressure candidates fall back on memorised lists, and SP8 questions often disguise a standard topic inside an unusual scenario, so the list feels relevant but earns little credit for application.
What to do instead
For each point, force a second sentence beginning 'In this case...' that names the specific exposure, dataset, or constraint. Before writing, spend two minutes highlighting the scenario-specific facts and plan to reference each one explicitly.
Ignoring the command word and mark allocation: writing an essay for a 'state' or 'list' question and only two lines for a 'discuss' or 'evaluate' question worth many marks.
Candidates do not calibrate length to marks, often because they have not practised reading the verb and the bracketed mark total together as a planning cue.
What to do instead
Use a rule of thumb of roughly one distinct creditworthy point per mark, and match depth to the verb (list = short points, discuss = point plus justification plus implication). In mock exams, annotate each question with the number of points you intend to make before writing.
Failing to show how a reserving or pricing result changes when a key assumption is varied, and not commenting on the sensitivity or the direction of the effect.
Candidates compute a central estimate and stop, because sensitivity analysis is often treated as an optional extra rather than a core part of the specialist judgement being tested.
What to do instead
Build a habit of adding a short sensitivity paragraph to every major calculation: pick two or three key assumptions (loss ratio, development factor, discount rate), state a plausible range, and describe the effect on the result and on the decision.
Confusing the purposes and mechanics of different reserving bases, for example mixing incurred but not reported estimates with case reserves, or applying a development factor to the wrong triangle direction.
The terminology overlaps and candidates learn definitions in isolation without mapping each method to its data requirements, its assumptions, and the line items it produces in the accounts.
What to do instead
Build a one-page comparison table for reserving methods (chain ladder, Bornhuetter-Ferguson, expected loss ratio, average cost per claim) covering data needed, key assumptions, strengths, weaknesses, and when each is appropriate. Redo past calculation questions from first principles rather than memorised templates.
Treating reinsurance as a single topic and not distinguishing proportional from non-proportional structures, or failing to explain how a given structure changes the cedant's retained risk, capital, and reserve volatility.
Candidates memorise definitions of quota share, surplus, excess of loss, and stop loss but do not practise applying them to a numerical example or explaining the economic trade-offs.
What to do instead
For each structure, write out the retained and ceded loss formulas, then work a small numerical example with a loss and an event limit. Practise explaining in two sentences how the structure affects capital requirements and earnings volatility for the cedant.
Describing the steps of a capital model or risk framework without linking them to the business decision, regulatory requirement, or governance outcome the question asks about.
Capital modelling content is dense and candidates reproduce process descriptions, missing that SP8 questions usually ask what the output is used for and how it should be governed.
What to do instead
For each modelling topic, prepare a short 'so what' note: which decision it supports, which stakeholders rely on it, what validation and documentation are needed, and what happens if it is wrong. Answer questions by starting from the decision and working back to the model.
Presenting only one option or recommendation in a question that asks for an evaluation, with no alternatives, no criteria, and no conclusion.
Candidates see a scenario and immediately argue for the first plausible approach, skipping the structured comparison that the higher marks reward.
What to do instead
Use an explicit structure: list the feasible options, state the criteria (cost, capital, risk transfer, regulatory acceptability, practicality), compare options against each criterion, then give a clear recommendation with the main caveat. Even a short answer should contain options, criteria, and a conclusion.
Misreading data given in the question, for example mixing cumulative and incremental claims, using earned rather than written premium, or ignoring the currency and time period stated.
Exam questions deliberately include realistic tables with several columns, and candidates under time pressure grab the first similar-looking figure without checking its definition.
What to do instead
Before calculating, write a one-line data dictionary for the question: what each column is, its unit, whether it is cumulative or incremental, and the period covered. Reconcile totals where possible as a check before using the data.
Running out of time because early questions are over-answered, leaving later high-mark questions with brief or missing answers.
Candidates do not rehearse time allocation per mark, and the temptation to perfect an early answer is strong when the first question feels familiar.
What to do instead
In mocks, allocate minutes in proportion to marks with a fixed reserve for review, and enforce a hard stop per question. If a question is unfinished, leave a short bulleted skeleton of the remaining points and move on, returning only if time allows.