In August 2019, the health actuarial unit at AXA Mansard Insurance changed hands. The person who took charge of it, Chinaza Ukatu, had walked into the company as an analyst fourteen months earlier and had left the University of Lagos only three years before that. Very little about the appointment followed the conventional path, least of all the route she had taken to reach it.

Ukatu studied Actuarial Science at University of Lagos, finishing in January 2016. It is a small discipline in Nigeria, and a demanding one, and the students who choose it tend to know early that they are heading for an insurer or a pension fund. She led the University of Lagos Student Actuaries as its president while she was still an undergraduate.

The department at Unilag graduates small classes, small enough that the people who teach there remember individual students and follow where they end up.

“We take in small numbers, so we know them well. Chinaza was one of the students who asked where a figure had come from rather than what she was supposed to do with it, which is the instinct the profession actually runs on. The complaint in this industry used to be that Nigeria had no actuaries. The complaint now is that the data is not ready for the ones we have, and the graduates who matter are the ones who go and fix that before they price anything.”

— Ashim Sogunro, Ph.D., Lecturer, Actuarial Science | University of Lagos

What she did not do was go straight into an actuarial department. In March 2016 she joined QuantumX in Lagos as a business analyst and spent the next thirteen months on cost benefit analysis, project evaluation and market research, much of it in the renewable energy sector, alongside work supporting mergers and acquisitions.

That detour turned out to matter more than a line on a curriculum vitae suggests. Renewable energy projects in Nigeria at that point had thin operating histories. There was no long series of local performance data to lean on, no settled benchmark for what a solar installation in a particular state would actually yield over a decade. An analyst asked to put a number on such a project has to build a view of risk out of adjacent evidence and stated assumptions rather than out of experience. That is, in substance, the actuarial problem, and she was doing it before she ever priced an insurance policy.

“You are taught to price off experience. In a sector with no local experience to price off, you have to be explicit about what you are assuming and where each assumption came from. I started writing the assumption set down as a separate document, so that anyone reviewing the number could see which part of it was data and which part was judgement. That habit turned out to be most of the job,” she says.

In May 2017 she moved to BukiHQ, also in Lagos, as a customer relationship manager. For a little over a year she worked the commercial side of the business: client reviews, relationship management, and the design of retail credit product combinations. It is an unusual year to find on an actuary’s record. Pricing is a back office discipline, and the people who do it rarely spend twelve months sitting in front of the customers who will eventually be quoted the numbers.

She returned to actuarial work in June 2018, joining AXA Mansard as a health actuarial analyst. The work was foundational rather than glamorous: data quality checks and validation on actuarial calculations, review of the health portfolio, valuations, sensitivity analysis and financial projections that fed management decision reports. She also worked alongside the product team on the design and pricing of the company’s health products, which is where the year at BukiHQ started to earn its keep.

“A lot of actuarial work fails at the validation stage rather than at the modelling stage. If the exposure data and the claims data disagree with each other, the elegance of what you build on top of them is irrelevant. So, the first question I ask of any figure is not whether the model is right, it is whether the inputs reconcile,” Ukatu says.

By that August, fourteen months after she joined, she was leading the unit. The Health Actuarial Unit at AXA Mansard carries the pricing and reserving work behind a health insurance book of roughly 53 million dollars serving more than a million customers, a book whose sales had grown 127 per cent year on year. Reserves are the least visible and least forgiving part of an insurer’s balance sheet. Set them wrong and nobody notices for a year or two, and then everybody does.

“Reserving is where an insurer eventually finds out whether it was right. You are estimating liabilities that will not settle for months or years, so the feedback arrives long after the decision that caused it. That is why the valuation work has to be conservative in method and transparent in assumption, and why I was not willing to sign anything off until the claims history underneath it had been cleaned,” she says.

Two pieces of work from that period give the shape of how she operates. The first was a repricing of the entire health product range, built off a large Microsoft Access claims database that had never been interrogated at that scale; in the period after the new prices took effect, sales rose about 10 per cent. The second was less visible and arguably more useful: she procured, cleaned and interpreted seven years of claims data in order to automate the health claims process in R and Excel, and cut a routine that had taken four hours down to about thirty minutes.

“The claims process was taking four hours because it was effectively being rebuilt by hand every cycle. Seven years of history were sitting in a database nobody had queried at that scale. Once the data was cleaned and the process was written as code in R, it ran in about thirty minutes, and more importantly it ran the same way every time. Reproducibility matters more than speed. A pricing decision you cannot reproduce is not a decision, it is an opinion,” she recalls.

Neither of those is the kind of achievement that produces a press release. Both are the kind that quietly changes what a company is able to do next.

“Health insurance in this country has never been short of data. It has been short of structured data, and short of people willing to spend months cleaning it before anybody gets to see a dashboard. Everything that came after that for me, the automation, the analytics, and now the research I do on artificial intelligence in financial reporting, sits on that unglamorous layer. Whatever the models look like in ten years, they will be worth precisely what the data underneath them is worth,” she says.

By the close of that year the unit she had inherited a few months earlier was running on cleaner data, a repriced product range and a claims process that no longer consumed half a working day. Within months, consulting rooms across Nigeria would begin closing, and AXA Mansard would need a price for a telemedicine product that no Nigerian insurer had any experience of selling. The groundwork for that answer had already been laid.