Actuary interview questions, with worked answers

How to use these questions

Many rounds include a case: a small data set to price, a claims triangle to reserve, or a result that moved and must be explained. Practise thinking aloud: say what you would check first, which assumption matters most and how sure you are. In the other questions, give a real piece of your work, your part in it and what happened next, without confidential figures.

What an actuary is expected to know

• The models: actuarial modellingBuild and run the cashflow models behind prices, reserves and capital, and know what each assumption does to the result. of cashflows for prices, reserves and capital.
• The work: pricingTurn claims, expenses and lapses into a price that covers the risk, and test it against the market., reservingEstimate what claims will still cost, including those not yet reported, and defend the estimate. and the capital rules, with IFRS 17Follow how insurance contracts are measured in the accounts and explain movements in the margins to finance. for the accounts.
• The numbers: statisticsChoose and test models for mortality, lapses or claims, and know the uncertainty around them. for assumptions and data analysisRead experience data for trends and errors before it goes into a model. before any model runs.
• The people: explaining resultsExplain a result, its assumptions and its uncertainty in plain words to people who are not actuaries. to finance, management and the supervisor.

The tools are named in ads side by side and none is a must everywhere: FIS Prophet and AXIS are common modelling tools for life business, and PythonClean and analyse data, automate runs and build small models with clear, tested scripts., RFit statistical models and produce reproducible analysis, common in pricing and reserving teams., SQLPull the policy and claims data you need from the company's databases, and check it adds up. and ExcelBuild clear, checked workbooks for smaller calculations and for the summaries others read. carry most of the data work. Interviewers want to hear how you checked a model's output, whatever the tool.

Pricing, reserving and assumptions

How would you approach the pricing of a new term life product?

What they check: the building blocks of a price and the judgement in each.

Short answer: "I project the expected claims from a mortality assumption, add the lapses, expenses and commission, discount the cashflows, and add margins for risk and profit. Then I test the price for the assumptions I am least sure of and compare it with the market."

The full answer: A lapse is a policy the customer stops paying for; lapses change both the profit and which customers stay. Mortality comes from industry tables adjusted to the company's own experience.

A weak answer: "I copy the competitors' rates." Follow-up: "Sales wants the price ten points lower." Say what you show them.

Claims on a motor portfolio are reported years after the accident. How do you set the reserves?

What they check: reserving methods and their limits.

Short answer: "I build triangles of paid and reported claims by accident year, project them with methods such as the chain ladder, and compare with a method that starts from an expected loss ratio for recent years. The reserve covers claims reported but not settled and IBNR, the claims incurred but not yet reported."

The full answer: Check what changed in the business, such as claims handling speed or a large claim, before trusting the pattern; a method only repeats the past.

A weak answer: "The chain ladder gives the answer." Follow-up: "Claims are now settled faster than before." Say what that does to your projection.

How do you set a lapse assumption, and when do you change it?

What they check: turning experience into an assumption.

Short answer: "From an experience study: I compare actual lapses with the expected ones by product, duration and channel over several years, and set a best estimate, the most likely value with no margin for prudence. I change it when the evidence is credible and lasting, not after one bad quarter."

The full answer: Write down why you changed or kept it; next year's reviewer and the auditor will ask.

A weak answer: "We keep last year's numbers." Follow-up: "Lapses doubled for one channel this year." Say what you look at first.

Your experience study shows mortality lighter than the assumption. What do you do?

What they check: credibility and the effect on each product.

Short answer: "I check how much data stands behind it and whether one group drives it, then blend our experience with the industry table by its credibility. Lighter mortality raises the cost of annuities and lowers that of life cover, so I show the effect on both."

The full answer: Also ask whether the trend will continue; improvements in mortality are an assumption of their own.

A weak answer: "Good news, we lower all reserves." Follow-up: "The study covers only three years." Say how that changes your answer.

Capital, accounts and reinsurance

What does a solvency capital requirement measure, and where does a stress test fit in?

What they check: the purpose of capital rules, beyond the formulas.

Short answer: "It is the capital an insurer needs to survive a very bad year, set by a standard formula or an internal model. A stress test shows what happens to the balance sheet in a named scenario, such as a market fall or a pandemic, so management sees where the firm is weak."

The full answer: Each place has its own rules for this: Solvency II in the EU, the NAIC's risk-based capital in the US, Solvency UK in the UK, the Swiss Solvency Test in Switzerland and IRDAI's rules in India. Each has its own reports, and this is not professional advice.

A weak answer: "Capital is whatever the model prints." Follow-up: "The ratio fell sharply this quarter." Say how you explain it.

Explain the contractual service margin under IFRS 17 to a finance director.

What they check: a hard idea in plain words.

Short answer: "The contractual service margin is the profit we expect from a group of contracts that we have not earned yet. It sits on the balance sheet and is released into profit as we provide the cover, year by year. Next to it, the risk margin, called the risk adjustment here, is what we hold for the uncertainty in the cashflows."

The full answer: Changes in assumptions about future service mostly move the margin, not this year's profit; that is often the surprise for finance.

A weak answer: A list of the standard's paragraphs. Follow-up: "Why did profit fall although sales grew?" Say what you look at.

How would you judge whether a reinsurance treaty is worth its cost?

What they check: reinsurance as a business decision.

Short answer: "I compare what we pay with the claims we expect to recover, and add what the treaty does for capital and for the volatility of results. A treaty can cost more than it recovers on average and still be worth it if it saves capital or protects against a large loss."

The full answer: Check the counterparty's strength and the treaty wording; recoveries depend on both.

A weak answer: "Reinsurance is always a cost." Follow-up: "The reinsurer raises the price by a third." Say what options you look at.

A company asks why its pension deficit grew although nothing changed in the scheme. What do you say?

What they check: the assumptions that move a pension valuation.

Short answer: "The deficit is the value of the promised pensions minus the assets. If market interest rates fall, the discount rate falls and the promises are worth more today; if assets did not rise as much, the deficit grows. I show the change split into its causes."

The full answer: Inflation and life expectancy move it too; a chart that bridges last year's deficit to this year's settles most questions.

A weak answer: "The model changed." Follow-up: "The finance director wants a lower discount rate assumption avoided." Say how you answer.

Models, data and checks

How do you check a model you did not build before you rely on it?

What they check: model validation and healthy doubt.

Short answer: "I read its documentation and assumptions, run simple cases by hand, compare results with last year and with a rough independent estimate, and test how it reacts to changes. Anything I cannot explain goes back to the owner before I use it."

The full answer: Validation is a documented step, often by someone outside the team; keep a record of what was tested.

A weak answer: "If it runs, it is right." Follow-up: "The model owner has left the company." Say what you do.

The data extract for the valuation has gaps, and the deadline is in three days. What do you do?

What they check: data quality under time pressure.

Short answer: "I measure the gaps first: how many records, which products, how much reserve they carry. Then I fix what I can at the source, use a documented approximation for the rest, and tell the sign-off actuary what I did and how much it could move the result."

The full answer: Reconcile the data with the accounts before anything else; a gap found after the run costs more time.

A weak answer: "I fill the gaps with averages and say nothing." Follow-up: "IT cannot fix the extract before the deadline." Say what you report.

Where do machine learning and AI fit into actuarial work?

What they check: a sober view of new tools.

Short answer: "They help most with large data: claims triage, fraud signals or finding the factors in pricing data. The result still has to be explained, validated and fair to customers, so I use them next to methods I can explain, not instead of them."

The full answer: AI assistants can write first drafts of code or reports, and a person checks every line; confidential data goes only into tools the company has approved.

A weak answer: "AI will replace actuarial models." Follow-up: "The model is accurate but nobody can explain a single price." Say what you do.

Explaining results and standing by them

How do you explain a reserve increase to people who are not actuaries?

What they check: clear communication of a technical result.

Short answer: "I start with the number and the main reason, then show a simple bridge from last quarter: new claims, changed assumptions, model changes. I say what is uncertain and what we would watch next."

The full answer: One page and one chart are often enough; the detail goes in an appendix.

A weak answer: "It is what the model says." Follow-up: "The board asks whether it will happen again." Say how you answer.

A manager wants a lower reserve to reach a profit target. How do you respond?

What they check: professional judgement under pressure.

Short answer: "I listen for any new facts, because those may change the estimate. If there are none, the reserve stays where the evidence puts it, and I explain why. The sign-off is a professional responsibility, and the actuarial bodies' standards back it."

The full answer: Write the discussion down and escalate to the head of the function if the pressure continues.

A weak answer: "I lower it a little to keep the peace." Follow-up: "The manager says other firms reserve less." Say what you check.

The regulator asks why your assumptions differ from last year's. How do you prepare?

What they check: documentation and calm.

Short answer: "I take the experience study, the decision record and the effect of each change on the results, and I explain them in that order. If the evidence is thin I say so and show the range we considered."

The full answer: A supervisor wants to see a process, not only an answer; show who reviewed and approved the change.

A weak answer: "We changed them because the results looked better." Follow-up: "Your assumption is outside the range of your peers." Say how you answer.

How do you balance the actuarial exams with a full workload?

What they check: planning and steady progress.

Short answer: "I agree study days with my manager in advance, plan the exam sittings around the valuation calendar, and study a little every week rather than in bursts. When I fail a paper I look at what went wrong and sit it again."

The full answer: The bodies differ by country: the DAV in Germany, the SOA and the CAS in the US, the IFoA in the UK, the Swiss SAV and the IAI in India each have their own syllabus.

A weak answer: "Exams come second." Follow-up: "The year-end valuation clashes with an exam sitting." Say how you plan.

Senior rounds

As a senior actuary or head of actuarial, how would you set up the peer review of a valuation?

What they check: a senior view of quality and ownership.

Short answer: "A peer review is an independent check by an actuary who did not do the work. I agree what is reviewed by risk: data, assumptions, model changes and results. The reviewer records findings, the team answers them, and only then do I sign off."

The full answer: Keep reviews proportionate: the large and the new get the most care, and actuarial analysts learn most from seeing the findings.

A weak answer: "I check everything myself the night before." Follow-up: "The reviewer finds an error a day before the board meeting." Say what you do.

Your CV for these questions

Each job on your CV invites the questions above: which business, which models, which results you owned and how you explained them. Give those stories a line each, written so a chief actuary wants to hear more:

Check your CV against an actuary job you want with the free score; any area it finds thin is the one to rehearse first, with a real piece of work told without confidential figures.