IAI CP1 common mistakes
IAI CP1 · Understand the traps and how to avoid them.
Writing the final answer as a single number (e.g. a reserve or premium) without showing the intermediate cash flows, discount factors, or commutation functions used to reach it, so marker-awarded method marks are lost even when the number is right.
Candidates coming from multiple-choice backgrounds treat CP1 as answer-only and skip the audit trail; under time pressure the working is the first thing sacrificed. They also assume a correct final figure automatically earns full credit.
What to do instead
Before the exam, build a habit of formatting every calculation as: state formula in symbols, substitute numbers, give intermediate values to 4 significant figures, then bold the final answer with units and timing (e.g. 'per annum, payable annually in advance'). Practise under timed conditions with a self-imposed rule that no numeric answer is written without at least two visible intermediate steps.
Producing a technically correct solution that never engages with the scenario's specific facts (named product, distribution channel, regulator, or policyholder group), so the answer reads as a generic textbook derivation and loses the application marks.
It is faster and feels safer to reproduce memorised theory; candidates fear that applying facts incorrectly will cost more than not applying them at all. Past-paper practice often rewards recall, which reinforces the behaviour.
What to do instead
Adopt a two-pass structure for every long question: first list the 3-5 scenario facts that change the answer (e.g. 'unit-linked', 'single premium', 'selling through independent brokers'), then explicitly reference each in the solution ('because the product is unit-linked, the investment risk sits with the policyholder, so...'). Do targeted drills where you write only the application sentences, not the theory.
Answering the verb that was not asked: giving a list of advantages when the question says 'discuss', giving a description when it says 'compare', or giving a recommendation when it says 'explain the limitations of the model'.
Under time pressure candidates skim the stem, see a familiar topic (e.g. 'unit pricing'), and dump everything they know about it. Command words look like decoration rather than the mark allocation.
What to do instead
Circle the command word and any plural nouns ('two reasons', 'three risks') before starting. Match structure to verb: 'compare' needs paired points with an explicit basis of comparison; 'discuss' needs both sides plus a judgement; 'calculate' needs method marks. After writing, re-read the stem and tick off each demanded element, deleting anything that does not serve it.
Ignoring the timing and payment-frequency conventions embedded in the question, for example discounting an annuity-due as if it were in arrears, or treating a premium as payable at the start of the year when the question implies otherwise.
Candidates pattern-match to the last similar question they practised and default to the convention they memorised. Small words in the stem ('in advance', 'immediately', 'first payment now') are easy to miss and change the answer by a full year of discounting.
What to do instead
Draw a small timeline for every cash-flow question, marking t=0, t=1, ... and placing each payment on it before touching a formula. Explicitly write the convention you are assuming in the answer ('premiums payable annually in advance, so a-double-dot'). Build a one-page personal checklist of conventions (annuity-due vs arrears, continuous vs discrete, immediate vs deferred) and run it at the start of each question.
Presenting a model or projection output as if it were the answer, with no comment on the reasonableness of assumptions, no sensitivity analysis, and no discussion of limitations.
Candidates believe the numerical output is the deliverable and that critique is 'extra'. They also lack confidence in judging assumptions, so they avoid the discussion to protect time.
What to do instead
Treat every modelling question as three deliverables: results, reasonableness check, and limitation. Check results against a simple independent estimate (e.g. compare a projected premium with a rough manual calculation or a known benchmark). Always add at least two sensitivities (e.g. mortality +/-10%, interest rate +/-1%) and state which assumption dominates the answer. Draft a reusable 'model critique' paragraph covering data quality, parameter uncertainty, model error, and communication to non-technical users.
Failing to state and justify the choice of basis (interest rate, mortality table, expense assumptions, margin for risk) and instead silently plugging in a single set of numbers.
Candidates see the basis as given data rather than a judgement they must defend. They also confuse 'best estimate' with 'prudent' and use whichever is easier to compute.
What to do instead
For every valuation or pricing question, write one sentence naming the basis and one sentence justifying it against the purpose (e.g. 'a prudent basis is appropriate because the reserve must be sufficient with high probability; a best-estimate basis is appropriate for an internal projection of expected profit'). Practise stating the direction of prudence ('higher mortality assumption increases the reserve for a term assurance').
Producing a recommendation or conclusion that does not follow from the analysis, or omitting a conclusion entirely because time ran out, leaving a list of points with no decision.
Candidates spend too long on calculation and treat the concluding paragraph as optional. Others fear committing to a judgement in case it is 'wrong', not realising that a reasoned judgement earns marks.
What to do instead
Budget the last 10% of time for conclusions and write them even if the calculation is incomplete: 'On the basis of the above, option A is preferable because X outweighs Y, subject to the caveat Z.' Practise writing one-sentence conclusions for every past-paper part, and make sure the conclusion names the criterion used (cost, risk, capital, policyholder outcomes).
Misreading which party bears which risk (e.g. attributing investment risk to the insurer on a unit-linked contract, or forgetting that a with-profits policy shares risk between policyholder and shareholder), leading to an answer that is internally inconsistent.
Risk transfer is taught as a list of product features rather than as a mapping from product design to who bears what. Under exam pressure the mapping collapses.
What to do instead
Build a one-page matrix: rows = products (term assurance, whole life, endowment, unit-linked, with-profits, annuity), columns = mortality/longevity, investment, expense, persistency, and who bears each. Rehearse it until you can reproduce it from memory in two minutes, then use it as the skeleton for any product-design or risk question.
Answering a professionalism or ethics scenario with a purely technical answer, or with a vague 'act with integrity', without identifying the specific parties affected, the specific guidance relevant, and the concrete course of action.
Candidates treat ethics as a soft topic and either skip it or write platitudes. They also do not know the structure expected: identify the issue, the parties, the relevant principle, the options, and the recommended action.
What to do instead
Use a fixed five-step template for every scenario: (1) what is the conflict or issue, (2) who is affected and what do they expect, (3) which principle or standard applies, (4) options with consequences, (5) recommended action including escalation to a senior actuary or the professional body. Practise on short scenarios until the template is automatic.
Poor time and paper management: spending disproportionate time on an early calculation, leaving later high-mark discussion parts unattempted, or writing long prose for low-mark parts and thin answers for high-mark parts.
Candidates allocate effort by comfort rather than by marks; calculations feel productive and discussion feels uncertain. Without a clock-based plan, the first question absorbs the surplus.
What to do instead
At the start, compute marks-per-minute (total marks divided by total minutes) and write a target finish time for each question on the paper. Cap any single calculation at its mark allocation; if stuck, write the method and move on. Do at least two full timed mocks and review the time log, not just the content, adjusting the plan for the next attempt.