Insurance underwriting versus actuarial internships: accounts, models and judgment

01Separate the job from the insurer’s name

An insurer can recruit underwriting, actuarial, claims, finance, operations and technology interns at the same time. Travelers lists these as separate internship paths. Finding an insurance-company name in a search result does not tell you which of those jobs is being offered. Start with the business line, customer and expected output.

BLS describes underwriters as evaluating applications and coverage decisions, while actuaries use mathematical and statistical analysis to study risk. Those are occupation-level descriptions, not a list of decisions a student is authorized to make. The internship should specify the work you support and who reviews it.

Insurance underwriting is also different from underwriting a commercial loan. A broker-facing sales role, a carrier underwriting team and an actuarial pricing team can work with related information but have different responsibilities. When a description combines them, ask which task will occupy most of the intern’s time.

Sources[1][2][3]

02Underwriting exercise: reconcile a contradictory submission

Original fictional brief: a manufacturer requests coverage. The main application lists one location, 18 employees and no recent incidents. An attached operations summary mentions a second warehouse, seasonal staff and a process change six months earlier. Your assignment is to produce a discrepancy log, not an insurance quote.

Treat both documents as evidence with a stated scope, not as automatically authoritative. The warehouse might be outside the requested coverage, or the application might be incomplete. Seasonal employees might be included in the headcount or excluded. You do not yet know. A useful analyst turns each discrepancy into a question whose answer would change the next step.

Write a short internal note that identifies the information still needed and the person who should review it. Avoid rating the business safe or unsafe from these three observations. You have neither the complete submission nor an approved decision framework.

03Actuarial exercise: separate frequency from severity

A second fictional dataset contains 1,000 identical one-year policies, 50 claims and an average claim cost of $2,000. Aggregate claim cost is $100,000. Claim count divided by policy count is 0.05 claims per policy, and claim cost divided by policy count is $100 per policy. These are descriptive calculations under deliberately simplified assumptions.

If claim count remains 50 and average severity rises 25% to $2,500, aggregate cost becomes $125,000, or $125 per policy. If severity stays $2,000 but claims rise to 60, cost becomes $120,000, or $120 per policy. The first change is driven by severity, the second by frequency.

The table varies both assumptions. Every cell is claim count multiplied by average claim cost, divided by 1,000 policies. Check the corners as well as the central case. The range is a scenario range, not a confidence interval.

04Explain why the answer is not a premium recommendation

The exercise excludes expenses, coverage terms, payment timing, capital, margins and differences among insured risks. It also assumes that the observed or hypothetical experience is representative of the future. A $125 claim-cost estimate therefore cannot simply become the recommended premium.

Write a limitations note organized by what you would need to estimate next. For example, you would need to understand whether claims are fully developed, whether exposure is measured consistently and whether the covered business changed. Do not fill those gaps by adding an unexplained percentage until the number looks commercially plausible.

For a portfolio, the defensible achievement is building and checking a frequency-severity sensitivity model. It is not pricing real policies, setting reserves or completing an actuarial opinion. The entire case can be done with the supplied synthetic inputs; private policyholder records add risk without improving this demonstration.

05Read exam requirements at the requisition level

Actuarial careers have a professional examination pathway, as BLS explains, but that does not make one particular exam a universal internship prerequisite. “Passed,” “scheduled,” “preferred” and “interest in pursuing” describe different states. Record an official pass only after the result is known, and identify a planned sitting as planned.

Read the degree field, graduation window, tools and location in the actual requisition. For underwriting, apply the same discipline to insurance credentials and licensing language. A credential advertised on a careers website is not necessarily required for the internship in front of you.

You do not need to buy a course to complete either exercise here. First inspect a few appropriate postings, identify which gaps actually repeat, and choose preparation that addresses the work. This avoids spending time on qualifications that belong to a different business line or a full-time development program.

Sources[3]

06Give the interviewer something specific to question

For underwriting, explain how you reconciled incomplete information and wrote a useful follow-up question. For actuarial work, explain your model, the meaning of its units and which assumption drives the result. Neither story needs an invented business-impact percentage.

A truthful example is: “Reconciled three inconsistencies in a fictional insurance submission and built a follow-up log.” Another is: “Tested nine claim-count and severity scenarios and documented why the output is not a premium.” These describe observable work. “Approved commercial accounts” and “set insurance rates” would describe authority the exercise did not provide.

Ask the team for an example of an intern deliverable and the review it receives. Then choose the path whose actual tasks you want to do repeatedly, while keeping alternatives open when your qualifications fit both.