Put an actuary in front of a cash flow model and they will, given time, figure out how to run it, update it, modify it and extract its results.  That’s because actuaries naturally tend to be problem solvers, regularly developing solutions from first principles.

The drawback to actuaries being excellent problem solvers is that, while they will almost always find a workable solution, it may not always be the best, most robust or most efficient solution for the problem at hand.  Traditional actuarial training isn’t designed to encompass robust systems thinking and being a proficient coder does not always translate into the ability to create and maintain an actuarial system.  The systems, IT, and computer science landscape is an independent realm of thinking, with its own best practices, conventions and principles, which may not come intuitively to anyone not trained in the field.

Hence the advent of the systems actuary, a term that’s been informally coined in the actuarial community to describe an actuary who, according to Google’s Gemini AI tool, “focuses on the development, maintenance, and optimisation of actuarial models and systems, ensuring accuracy and reliability, while also providing technical support and collaborating with IT professionals to meet business needs.”

This definition broadly captures what a systems actuary does, but not what differentiates an actuary with bona fide systems training from one who has learnt what they know about systems by being thrown in the deep end and figuring it out on the job.

In this article, we explore why rigour and discipline around actuarial systems are important and describe some properties of a good actuarial system as identified by our computer science-trained systems actuaries.

The importance of discipline and rigour in building and maintaining actuarial systems

Imperfect actuarial systems do not always announce themselves as $6 billion London Whales.  They manifest as staff working overtime due to models crashing, late submission of results because of long model run times and audit findings when teams are not able to reproduce or explain modelled metrics.

The implementation of IFRS 17 brought systems issues – such as inadequate data storage, clunky data flow and an inability to integrate into other systems and business units – to the forefront of many actuaries’ minds.  Many of the flaws the actuarial systems could have originated from a lack of IT expertise at inception and the developments bolted on along the way.

Well-designed actuarial systems save costs in the long run by being more efficient, automated and robust.  A standardised approach makes the systems easier for new staff to understand and reduces key-person dependencies.  Reputational, model and compliance risks are also reduced as models are more clearly defined and run better and more predictably, with fewer errors and crashes.

To achieve such a well-designed system, it is essential that actuaries receive IT or computer science training or that they outsource systems work to actuaries formally trained in systems.

What makes a good actuarial system?

For any actuary interested in formalising their IT knowledge, researching, internalising and applying the following features and aspects of good systems thinking and systems actuary expertise may provide a good starting point:

End-to-end flow of data and results:

Ensures that actuarial calculations are traceable from raw data to final outputs, reducing errors and improving transparency in decision-making. Robust automated logging is of great use here.

End-user experience:

Affects how actuaries interact with models and tools, ensuring usability, accuracy, and efficiency in deriving insights from complex data.

Version control and auditability:

Essential for tracking changes in actuarial models, preventing unintended modifications, and enabling rollback to previous versions if needed, e.g. during audits.  This helps ensure compliance with governance standards and mitigation of risks from unauthorised modifications. Version control in actuarial modelling should arguably have a dedicated team/resource.

Backwards compatibility:

Ensures that legacy actuarial models and historical data remain functional and usable when systems are updated, avoiding disruptions.

Modularity:

Facilitates the development of reusable model and system components, allowing for flexibility and scalability in system enhancements, as well as quick development turnaround times for new features and products.

Extensibility:

Allows actuarial systems to adapt to future regulatory changes and business needs without requiring a complete system overhaul.

Performance and Efficiency:

Critical for efficiently handling large datasets and complex calculations, ensuring timely reporting and decision-making.

Time complexity:

Influences the feasibility of running actuarial models in real-time, affecting responsiveness and computational resource allocation. Computer science knowledge of time complexity of algorithms and how to assess them is crucial in making implementation decisions in more complex non-linear actuarial models.

Data and storage management:

Ensures actuarial systems maintain high data quality, integrity, and accessibility for accurate modelling and reporting, as well as a robust and efficient storage setup. Knowledge of relational databases and normalisation will help the actuary decide on how best to set up valuation data, assumptions, and results databases.

Audit logging and traceability:

Provides a clear record of changes and decisions in actuarial processes, essential for compliance and risk management.

Interaction with existing systems:

Ensures smooth integration with finance, risk, and underwriting platforms, reducing inefficiencies and data silos. Acknowledging that the software landscape is constantly evolving in response to changing reporting and business requirements, building systems that plug into other software applications easily and aren’t platform-locked, is of great value.

Appropriate selection of new tools and software:

Being aware of the latest trends and releases in the actuarial systems market allows the systems actuary to select the best solutions for their teams.

Optimisation of existing tools and software:

Enhances actuarial productivity by leveraging the team’s existing software toolkit, reducing unnecessary expenditure on new systems and software. The actuary should know some simple techniques for identifying bottlenecks in the system.

Fit for purpose:

Ensures that actuarial tools align with business and regulatory requirements, preventing inefficiencies and inaccuracies. The size and complexity of the implementation and tool should be appropriate for the particular business case.

Efficient coding/uses IT convention for system design:

Supports maintainability, scalability, and collaboration between actuaries and IT teams, reducing technical debt. Learning things like Boolean algebra, Backus-Naur Form, REST API use, git, and how to work with an Integrated Development Environment (IDE) will be of great value here.

Collaboration with IT:

Working closely with IT professionals to develop and integrate actuarial systems helps ensure that they meet business requirements and regulatory standards.  Systems actuaries are able to speak “IT”, which allows them to draft IT requirement specs in the correct language for BI and IT teams, acting as a communication bridge between Actuarial and IT.

How Insight can help

The Insight Life Solutions has a strong systems focus, and our actuaries approach systems problems with a passionate curiosity, backed by computer science qualifications, IT training and knowledge of the latest trends and players in the actuarial systems industry.

We are fluent in all the major actuarial systems and coding languages used in South Africa, so please be in touch with any actuarial systems queries, including:

• Model development
• Model migration, validation and review
• Systems design and development
• Systems optimisation

Pamela Hellig, Head of Insight Life Solutions (pamelah@insight.co.za)
Garrit Nieuwoudt, Head of Actuarial Systems at Insight Life Solutions (garritn@insight.co.za)