IAI SA3 cheat sheet

SA3 is an IAI (Institute of Actuaries of India) specialist-level paper; its syllabus centres on applying actuarial judgement to a defined specialist practice area rather than on generic foundation material.

Exam technique: answers are marked on demonstrated application of specialist concepts to the scenario given, so each answer must state the principle, apply it to the facts, and state a conclusion.

Regulatory context for Indian actuarial work is set by IRDAI (Insurance Regulatory and Development Authority of India); cite the relevant IRDAI regulation or circular where a question turns on compliance.

Actuarial work in India is governed by the Actuaries Act 2006 and the Institute of Actuaries of India; professional-conduct obligations (integrity, competence, confidentiality) apply to every answer involving advice to a client.

A specialist answer distinguishes statutory/regulatory requirements from professional judgement: state what the regulation mandates, then what the actuary recommends beyond the minimum.

Where a question asks for a recommendation, structure it as: objective, options considered, criteria, evaluation of each option, recommendation, and limitations/caveats.

Quantitative specialist questions usually require stated assumptions; marks are earned for defensible assumptions and sensitivity analysis, not only for the final number.

If a scenario is under-specified, state the missing information explicitly and proceed on a stated assumption rather than refusing to answer.

Model answers should identify the data required, the data-quality checks (completeness, consistency, credibility), and the treatment of missing or anomalous data before any calculation.

For reserving-type questions: distinguish case estimates from IBNR, state the method (e.g. chain-ladder, Bornhuetter-Ferguson, expected-loss), and justify the choice against the data volume and stability.

For pricing-type questions: separate expected cost (frequency x severity or loss ratio), expenses, profit/risk margin, and the effect of reinsurance; show the gross-to-net bridge.

For capital/solvency questions: distinguish regulatory capital from economic capital, and note that diversification benefits must be justified by evidence of non-perfect correlation.

For risk-management questions: use a standard risk taxonomy (insurance, market, credit, liquidity, operational, strategic) and map each identified risk to a mitigation and an owner.

Communication marks: define technical terms on first use, avoid unexplained jargon, and state conclusions in plain language suitable for a board or client audience.

Ethical/professional scenarios: identify the conflict, the affected parties, the relevant professional standard, the disclosure required, and the escalation route.

Time management: allocate marks proportionally to marks available; a 20-mark question warrants substantially more depth than a 5-mark question, not just more words.

Reinsurance decision rule: cede risk when the cost of reinsurance is below the risk-adjusted value of the capital and volatility it releases; retain when the loading exceeds that value.

Experience-rating and credibility: the weight given to own experience increases with volume; state the credibility basis (e.g. limited-fluctuation or Bühlmann) when blending own and manual rates.

Discounting: reserves and liabilities should be discounted only with a rate consistent with the currency, duration, and risk of the cash flows; state whether the rate is risk-free or risk-adjusted.

Sensitivity and scenario testing: a single central estimate is insufficient for a specialist answer; include at least one adverse scenario and state the impact on the result.

Data credibility: small or immature portfolios require pooling, external benchmarks, or wider margins; state which was used and why.

Documentation: every specialist recommendation should record data sources, assumptions, method, results, and the reviewer, so it can be independently reproduced.

Peer review and sign-off: material actuarial judgements affecting policyholder or shareholder funds require independent review before issue.

When a numerical answer depends on a judgement (e.g. a margin), present a range and state the point estimate chosen and the reason.

IFRS 17 relevance: under IFRS 17 insurers hold a Liability for Incurred Claims, a Liability for Remaining Coverage, and a Risk Adjustment; the risk adjustment reflects the compensation required for uncertainty, not a prudential margin.

IFRS 17 contract boundaries and onerous-contract tests change the timing of profit recognition; a specialist answer must state which measurement model (GMM, PAA, VFA) applies and why.

Indian solvency: IRDAI prescribes a control level of solvency margin (assets over liabilities, expressed as a ratio); answers should state the required level and the actual level, not just 'adequate'.

Embedded value reporting: distinguish the present value of in-force business, the required capital, and the cost of capital; free surplus plus required capital plus VIF equals embedded value.

Profit testing: project premiums, claims, expenses, reserves, and tax; discount at a rate reflecting the shareholder's required return; report IRR and NPV, not just one.

Asset-liability management: match duration and currency where possible; state the interest-rate and inflation sensitivities of both assets and liabilities.

Treaty and regulatory change: when a rule changes, quantify the effect on reserves, capital, and pricing, and state the transition treatment.

Tax: distinguish policyholder tax from shareholder tax, and state whether figures are gross or net of tax.

Claims reserving diagnostics: compare actual vs expected development each period; a persistent one-way deviation signals a change in settlement patterns or case-reserving practice.

Large-loss treatment: separate large and attritional claims; a single large loss can distort a loss ratio, so state whether it is included and its effect.

Expense analysis: split acquisition, maintenance, and claims-handling expenses; only expenses genuinely attributable to the contract should be loaded into its price.

Capital allocation: allocate diversified capital to lines using a marginal or Euler method; note that the allocation depends on the chosen method and is not unique.

Scenario design: choose scenarios that stress the specific dependencies in the portfolio (e.g. inflation plus claim frequency), not generic market shocks.

Back-testing: compare prior projections with emerging experience and state whether the model was validated, recalibrated, or rejected.

Uncertainty communication: distinguish parameter uncertainty, model uncertainty, and process (random) uncertainty; they are managed differently.

Recommendation quality: a recommendation without stated limitations is incomplete; always state what could make the recommendation wrong.

Stakeholder analysis: identify policyholders, shareholders, regulators, reinsurers, and distributors; a recommendation that favours one at the expense of another must say so.

Fair treatment of customers: pricing and reserving decisions must be defensible to the regulator as fair; discriminatory or opaque treatment requires explicit justification.

Disclosure: if a result is sensitive to an assumption the client has not seen, disclose the assumption and the sensitivity, not just the headline figure.

Conflict of interest: declare it, and if it cannot be managed, decline the engagement or escalate to the professional body.

Whistleblowing: where a legal or regulatory breach is suspected, the actuary's duty to report to the appropriate authority overrides client confidentiality.

Competence boundary: if a question falls outside the actuary's expertise, state the need for specialist advice (legal, tax, IT) rather than guessing.

Record-keeping: retain the working papers and the basis of the advice for the period required by professional standards and regulation.

Review cycle: state when the analysis should be refreshed (e.g. annually, on a material change in experience, or on a regulatory change).