IFoA SA4 common mistakes
IFoA SA4 · Understand the traps and how to avoid them.
Spending the first 20-30 minutes writing a long, unstructured list of 'all the things I know about reserving' before deciding what the question actually asks, so the answer opens with background and the actual judgement is squeezed into the final paragraph.
Candidates trained on calculation-heavy exams default to dumping knowledge because it feels productive, and SA4's open-ended 'discuss' wording gives no obvious stopping point, so the brain keeps generating content instead of committing to a position.
What to do instead
Before writing, spend 3-5 minutes producing a one-line 'answer in a sentence' (the recommendation or conclusion) plus a bullet skeleton of the 4-6 points that support it, then write to that skeleton. If a sentence does not support the stated conclusion or address a named stakeholder, cut it. Reserve the first paragraph for the conclusion, not the background.
Answering a stakeholder question in the wrong register: giving a board a page of technical caveats and model mechanics, or giving the actuarial function a hand-wavy 'it should be fine' with no numbers, assumptions or sensitivity.
Candidates write for the examiner-as-marker rather than for the named audience, and under time pressure they reproduce whatever is easiest to write rather than adapting content, structure and vocabulary to the reader's technical level and decision needs.
What to do instead
Identify the audience in the question and apply a fixed test: for a board, lead with the decision, the range, the key risk and what you recommend, pushing method to an appendix; for technical peers, state assumptions, data, method, results and limitations explicitly. Practise rewriting one answer twice for two audiences and compare what you deleted.
Treating a 'recommend' or 'advise' question as a neutral textbook essay: listing pros and cons symmetrically and never committing to a recommendation, or committing without stating the criteria and trade-offs used to choose.
Candidates fear that a firm recommendation will be 'wrong' and lose marks, so they hedge; but SA4 marks reward a defensible, justified judgement, and the hedging itself signals weak application of judgement.
What to do instead
State the decision criteria first (capital efficiency, policyholder outcomes, regulatory acceptability, cost, timescale), score the options against them in a short table or structured list, then give a clear recommendation with the main condition under which you would change it. Accept that a well-justified recommendation scores better than a balanced non-answer.
Ignoring the specific company context in the scenario (its size, mix of business, reinsurance structure, capital position, regulatory regime, distribution channel) and writing generic answers that would apply to any insurer.
Under time pressure candidates pattern-match to a remembered topic and reproduce standard lists, missing that the scenario details are deliberately included as the discriminators the examiner wants used.
What to do instead
Before answering, underline every quantitative or structural fact in the scenario and force at least one explicit link to each in the answer ('given the 60% motor concentration and the 2024 reserve strengthening...'). If a generic point cannot be tied to a scenario fact, replace it with one that can.
Producing numbers without a stated basis: quoting a reserve estimate, capital requirement or reinsurance cost with no assumption, method, data source, uncertainty range or sensitivity, so the figure looks authoritative but is unsupported.
Candidates want to demonstrate technical competence by producing a number quickly, and the exam format rewards visible output, so the underlying assumptions and their sensitivity get skipped.
What to do instead
For every number, write the basis in the same sentence or immediately after: method, key assumptions, data used, and a range or sensitivity to the two assumptions that matter most. Where data are inadequate, say so and state what you would do (proxy, expert judgement, wider range) rather than silently assuming.
Treating reinsurance purely as a cost line, ignoring its interaction with capital, solvency coverage, earnings volatility, counterparty credit risk, and the effect of reinstatements and aggregate covers on the tail.
Reinsurance is often learned as a pricing and coverage topic, so candidates default to 'buy more to reduce volatility' without modelling the net-of-reinsurance capital position, the cost of capital released, or the residual risk retained.
What to do instead
For any reinsurance question, structure the answer around the gross-to-net bridge: retained versus ceded loss distribution, capital before and after, cost of cover versus capital released, counterparty exposure and reinstatement/aggregate limits. Always ask 'what does this do to the tail and to the solvency ratio', not just to expected cost.
Describing the regulatory framework from memory in vague terms ('Solvency II requires firms to hold enough capital') without engaging with the specific regime, the calculation approach, the use test, or how the regime shapes the commercial decision in the scenario.
Regulatory knowledge is broad and candidates revise it as a list of pillar headings, so they can name the framework but cannot apply its mechanics to a pricing, reserving or capital decision.
What to do instead
Learn each regime through three lenses: what it requires numerically (valuation basis, capital formula, risk margin), what governance it imposes (board responsibility, model validation, use test, ORSA), and what commercial behaviour it drives (product design, reinsurance purchase, capital allocation). Practise applying each lens to a short scenario in under five minutes.
Writing about enterprise risk management as a list of risk categories and committee names, without linking risks to quantified impact, appetite statements, controls, escalation triggers or the insurer's strategy.
ERM is easy to revise as vocabulary (risk register, three lines of defence, risk appetite) and candidates reproduce the vocabulary rather than demonstrating how risk management changes a decision.
What to do instead
For any ERM answer, pick the two or three risks that actually matter for the scenario and for each state the appetite limit, the exposure relative to that limit, the control, the monitoring metric and the escalation trigger. Show the feedback loop into strategy or capital, not just the existence of a framework.
Failing to address professionalism and the Actuaries' Code explicitly when the scenario contains an ethical pressure point (a director pushing a lower reserve, a client wanting a favourable opinion, a conflict of interest, reliance on unaudited data), or mentioning the Code only as a closing sentence.
Candidates treat professionalism as a separate revision topic rather than an embedded part of every technical judgement, and they assume the technical answer is what is being marked.
What to do instead
Scan the scenario for the pressure point and address it head-on: identify the threat (self-interest, advocacy, familiarity, intimidation), the relevant Code principles (integrity, competence and care, objectivity, communication, professional obligations), the action you would take and to whom you would escalate. Put this in the body of the answer, not a footnote.
Running out of time because early answers are over-written, leaving the final high-mark 'evaluate' or 'advise' question answered in bullet fragments, or ignoring the mark allocation and giving a 5-mark question a 20-mark answer.
Candidates write until the knowledge runs out rather than to the marks available, and open-ended SA4 questions have no natural stopping point, so time disappears on the first two questions.
What to do instead
Convert marks to minutes at the start (roughly one minute per mark plus reading time), write the allocation on the page, and move on when the time for a question expires even if imperfect. Practise full papers under timed conditions and review where you over-ran; deliberately write shorter answers to low-mark parts to bank time for the final judgement question.