Actuaries, Mathematicians and Statisticians
Analyse, develop and apply actuarial, mathematical, statistical, financial and quantitative principles and techniques to solve problems and design experiments and surveys in a range of fields such as business and finance, health, engineering, government, scientific and social research, and technology.
The number for tomorrow's board
You are an actuary at an insurance company, preparing an estimate of future claims for a board meeting tomorrow. It is late afternoon. The board meets in the morning and wants a single figure for expected claims. Your model is almost finished, but a few things do not sit right, and people want answers.
Choose what you would do. Then see what happens. There are often good reasons for each choice.
The board hates ranges. Give me one number for next year's claims and I'll present it with confidence. I don't need the uncertainty, just the figure. Can you do that?
Give one figure only, because that is what he asked for and the board will find it much easier to follow
The board treats the figure as certain. When the year turns out differently, they feel misled. Your number was real, but the uncertainty was part of the truth.
Give a best estimate with a simple range, and explain in a sentence why the range matters
Oliver sees that he can present the best estimate clearly and still be honest. The board asks better questions as a result. Plain honesty about uncertainty is part of the craft.
Refuse to give a single figure at all, since any one number is misleading
You are right that one number can mislead, but Oliver is left with nothing usable. Being helpful means finding a way to be both clear and truthful.
I've found something odd. A batch of last year's claims has no date in the data, and I'm not sure how our model handled them. Should we fix it now or flag it after the meeting?
Ignore it for now since the effect is probably small and the meeting is tomorrow, and you can sort it out later
The effect is not small. The missing dates were concentrated in one month, which skews the pattern. A shortcut under pressure can hide the very thing you are trying to measure.
Check how the model treated them, test the result both ways and note the assumption in your report
You find the effect is modest but real, and you show both versions. The board sees your confidence and your caution together. Mei learns that data problems are worth finding.
Delay the report until the data is perfectly clean, whatever the board says
You protect the quality of the data, but the board meets without your analysis. Waiting for perfect can mean missing the moment when your advice is needed.
I'm not sure I follow how you got this result. Can you explain it simply? And be honest, how sure are you? I'd rather hear doubts now than surprises next year.
Explain it in plain words, say what you are confident about and what you are not, and give an example
Dr Whitaker follows the main idea and appreciates the honesty. A simple picture or comparison would have helped even more with the harder parts.
Show the full technical method so that he can see it is thorough
The room's eyes glaze over. You have shown the work, but not the point. If people cannot follow what you tell them, they cannot make good decisions with it.
Use a simple everyday comparison, give your honest level of confidence, and invite follow up questions
Dr Whitaker nods and says that was the first time the model made sense to him. The board trusts a person who can explain the work and own its limits.
How it went
The board leaves with a clear number, a clear range and an honest explanation of both. Being an actuary means being comfortable with uncertainty and brave enough to say so, in language others can use. Does that sound like a good use of a careful mind?
This is a simplified, fictional situation to help you explore. It is not professional instruction or a test of your ability.
Case study: The trend nobody wanted to see
A fictional case set in the analytics team of a fictional general insurance company.
The situation
Wen was an actuary at Coastline Mutual, an insurance company. While reviewing claims from storms, she noticed that damage in a particular coastal area had been growing faster than the company's pricing model expected. The pattern was small but steady across several years. Her manager was proud of the model, and the quarterly results were due. If she raised the issue, it would mean more work, an awkward conversation with the board and possibly a drop in the company's expected profit.
What they did
Wen double checked her own work first. She tested the data in different ways and asked a colleague to review it. When the pattern still held, she took it to her manager with a short, clear explanation and three possible responses. She presented it to the board in plain words, making clear what she knew, what she did not and how uncertain the estimate was. She avoided scaring them or sounding certain beyond the evidence.
How it turned out
The company adjusted how it planned for that region and set aside extra reserves. The next year's storms hit that area hard. The board remembered the warning, and Wen was asked to help redesign how the model was reviewed each year.
What would you do?
- Why did Wen check her own work before telling anyone?
- How would you explain uncertainty to people who want a firm answer?
- What do you think it takes to deliver unwelcome news to a board?