IYMC Logo

What Does an Actuary Do With Insurance Data?

By Julia Berens · · 5 min read

Close-up of a weathered, scratched surface with golden-brown patches, dark streaks, and a pale silvery highlight near the top.

If you like probability, you might wonder: what does an actuary do with it at work? Imagine an insurer that covers bicycles against theft. To estimate what the coverage will cost, it needs to know both how often it will pay a claim and how much it will pay when one comes in. An actuary uses data to estimate both, checks what could make those estimates wrong, and explains what the results mean for the insurer.

What does an actuary do with past claims?

Suppose an insurer has records for 1,000 bikes, each insured for one full year under the same terms. In this made-up example, 40 bikes had a paid theft claim, and each of those bikes had just one. The other 960 had no paid theft claim. The observed share with a paid claim is 40 ÷ 1,000 = 0.04, or 4%. If next year's bikes and conditions are similar, 4% is a starting estimate for the chance that a policy has a paid theft claim during the year.

Keep track of what the numerator actually counts. The insurer is counting paid claims, not all thefts. An owner might not report a theft, or a loss might not be covered under the policy. If the insurer changes what the policy covers, the old paid-claim rate may no longer be a good guide, even if bikes are stolen at the same rate.

The denominator deserves attention too. We made the example easy by giving every bike a full year of coverage. If many bikes joined halfway through the year, each would have had only half a year in which to generate a claim in the insurer's records. Treating all those bikes as if they had a full year of coverage would distort an estimate of the annual claim rate.

Estimating the cost of a paid claim

Suppose the insurer paid $24,000 across the 40 claims. The average payout was $24,000 ÷ 40 = $600. Notice the change in denominator: to estimate the average payout when a claim is paid, divide by 40. Dividing by 1,000 would give the average payout across all insured bikes, including those with no claim.

Now use those two estimates for next year. Assume again that a policy can have at most one paid theft claim in the year. With a 4% chance of a claim and an average payout of $600 if there is one, the expected claim cost per policy is:

0.04 × $600 + 0.96 × $0 = $24

The $24 is an average predicted by the model, not a forecast that any one owner will receive $24. It is also not the premium, the price the customer pays for coverage. An insurer also needs to account for expenses such as processing claims, and it must consider whether it can handle years when payouts are higher than expected.

Checking what the average hides

Say the insurer expects to cover more expensive bikes next year. If the average paid claim rises from $600 to $750 while the chance stays at 4%, the expected claim cost rises to $30. If the average payout stays at $600 but the chance rises to 5%, the expected cost also rises to $30. Those two results look identical on a spreadsheet, but the explanations differ: one points to larger payouts, the other to more claims.

An actuary would check whether the data support either change. For example, a higher share of expensive bikes does not necessarily mean payouts will rise by the same amount as bike prices. A policy limit can cap what the insurer pays. An actuary can compare results across groups of bikes or policy types, while being careful not to read too much into a group with only a few claims.

Claims can also occur together. If several insured bikes are stolen from the same storage room, the insurer could face several payouts at once. That does not invalidate the expected cost per policy, but it matters when estimating the chance of an unusually expensive year.

High school math you can use now

The bike example uses percentages, averages, and a weighted sum. The word conditional describes the average payout because it applies only under a condition: a claim was paid. Algebra lets you run the calculation backward. If the expected claim cost is $30 and the average paid claim is $600, then solving p × $600 = $30 gives p = 0.05.

Try putting the example in a spreadsheet. Give the claim probability and the average payout their own cells, then make a third cell multiply them. Change the probability from 4% to 5% and the average payout from $600 to $750 at the same time. Before checking the result, write down what you expect. The answer is $37.50, not $36: when both inputs change, the higher claim rate applies to the higher payout. For more practice explaining the reasoning behind calculations, use the IYMC training and practice problems.

College majors and professional exams

Actuarial science is one college option, but it is not the only one. Mathematics or statistics can build the skills you used in the bike example, especially if you also take courses in economics, finance, and computing. Compare course lists rather than relying on the major's name. Look for probability and statistics, and for chances to work with real data and explain what your calculations do and do not show. The IYMC Studies and Careers articles offer other ways to explore where a math degree can lead.

Professional qualification is separate from choosing a major, and the route depends on where you plan to work. In the United States, the Bureau of Labor Statistics profile of actuaries describes a path that typically includes a bachelor's degree and a series of professional exams. For an example of exam content, see the Society of Actuaries Probability exam. Its material goes beyond taking a percentage of past claims: it includes probability distributions and expected values.

If you're comparing colleges, pick two programs and mark which courses teach probability, statistics, and programming. Then check the professional requirements where you might work. That gives you a concrete list of questions to ask a college adviser, such as which course covers conditional probability and when you would be ready for it.

Frequently asked questions

Is expected claim cost the same as the premium?

No. Expected claim cost estimates payouts. The premium is what the customer pays and also needs to account for costs such as administering policies.

Do I have to major in actuarial science?

No. Mathematics and statistics are also possible routes. Check that your courses prepare you for the professional requirements where you plan to work.

Can I practice actuarial math in high school?

Yes. Start by calculating claim rates and expected payouts in a spreadsheet, then explain how changing one assumption affects your result.