Over the last year or so, we’ve seen a pronounced trend of life insurers starting to reassess their actuarial systems landscapes and explore the possibilities of modernisation, specifically in the realm of cash flow modelling engines. In our industry-wide survey conducted in Q3 last year, more than 80% of participants said they were contemplating a change in their actuarial software landscape within the next five years.

What makes a modelling system desirable?

It is no coincidence that such introspection has come at the tail-end of vast IFRS 17 implementation programmes, which laid bare the inefficiencies and inflexibility that have been baked into antiquated legacy systems. Combine this realisation with the budget cuts that seem to be sweeping through the industry and a growing awareness of the potentially more suitable software options in the market and you have a recipe for an actuarial systems revolution.

It may, however, be a revolution of the slow and steady kind – embarking on a systems migration or modernisation programme has the potential to be a huge undertaking. Initial steps may include preparing a business case, conducting vendor trials, and managing an RFP process… that’s all before you start the implementation itself. But for those actuarial teams who have started contemplating the possibility of reassessing their cash flow modelling systems – a major step in itself – we have listed, from a systems actuary’s point of view, some desirable features of modern actuarial projection software that might be considered when evaluating providers:

Interoperability

• This enables seamless integration with existing systems and tools, allowing actuaries to build custom workflows without being locked into a single platform.
• Software that offers an Application Programming Interface (API) to execute models and consume results from external programming interfaces is beneficial.

Modularity

• This encourages reuse and maintainability by letting users swap or upgrade individual components without affecting the entire system. It also reduces scope for code errors and turnaround time for new product and feature implementation.
• Features like inheritance, multiple inheritance and interfaces are important for a modular system.
• The ability to write reusable classes and functions may also be desirable.

Orchestration functionality

• This allows the coordination of complex tasks across multiple modules or systems, ensuring the efficient execution of multi-step reporting processes.
• Some software will offer orchestration tools that work with their software and others, often at an additional fee.
• Having an API available may allow the use of third-party orchestration tools.

Fair pricing

• Offering cost-effective licensing makes systems accessible for teams of various sizes without sacrificing functionality.
• We believe that a concurrent user licence model is more desirable than a named user model.

Auditability

• This refers to maintaining transparent records of changes, inputs, and assumptions, which supports compliance and simplifies external or internal audits.
• Auditability also includes detailed run logging and runtime messages for debugging and early detection of errors.

Familiar coding language (e.g. Python, VBA)

• This lowers the learning curve, increases adoption by allowing actuaries to work in languages they already know, and makes recruitment of systems actuaries and actuarial modellers easier.

Use of modern coding frameworks (e.g. loops, classes, logic flow)

• This enhances code readability, reusability, and efficiency, leading to faster development and easier maintenance.
• Some software solutions offer simple logical flow in each variable, while others allow a fully-fledged script with any amount of logic to be included in each variable, including loops, classes, and sub-variable definitions. This can be powerful because not only does it allow for complex logic and calculation, but it can actually make such implementation code less complex and easier to follow compared to implementing the same thing in a more constrained system. An example would be to compare the ease of following a ten-line Python script versus the difficulty of following a three-line Excel formula.

Detailed help file

• This reduces onboarding time and supports troubleshooting with thorough, easy-to-navigate documentation.
• Software providers that enable their users to help themselves by providing detailed help files and examples should not be taken for granted.

Input and output in any format (ODBC)

• This increases flexibility by allowing direct connectivity to various data sources and formats, facilitating smooth data exchange.
• The ability to read and write to ODBC is desirable. At the very least, this ensures the model will be able to read data and write results from and to most database engines, such as MSSQL.

Text-based code files for better version control (using tools like Git)

• This enables robust version control, collaboration, and rollback capability through integration with standard development tools.

Built-in access control and change logging

• This enhances security and governance by tracking user actions and restricting access based on roles.
• The ability to assign some users as read/run only and some users as developers may be useful for larger teams.

Parallel processing and high-performance computing

• This is the ability to run models making use of multiple CPU cores, different computers, or scalable cloud computing environments such that run times scale with increased computing power.
• Some solutions can perform GPU modelling, allowing for extremely fast runs and enabling real stochastic modelling.

Debugging and Visual Analysis

• The ability to set a breakpoint in a model and then step through is very valuable.
• A software solution should enable some ability to visualise a projection for a single model point showing variable precedents and dependents and their values during a test projection. We would consider this a must-have feature.

How do we know which system is right for us?

The features listed above provide a general guide to what makes a projection system desirable, but they may not all be necessary or suitable for every insurer. When assisting clients in assessing software options, we tend to evaluate them based on the following categories:

Learning Curve

Taking on a newer, cutting-edge solution means there is less availability of adequately skilled resources in the market. For some platforms, the learning curve is so flat that this isn’t a problem, and self-teaching is possible. In other cases, the lack of expertise in the market can render the platform non-viable.

Features vs. Price

How much do you get as a standard, what would need to be paid for separately, and how much do the extras cost? The decision around which system to go for will depend heavily on whether the features important to one’s business are included in the standard package or are at least affordable extras.

Ease of Implementation

As in programming languages, there is generally a trade-off when it comes to the power vs. speed of implementation for actuarial projection systems. A higher-level language allows for fast implementation turnaround, but it may not run as efficiently or be as robust. Similarly, with actuarial software, a smaller company with a single book of business may choose a lightweight, easy-to-manage solution over one that is more complex and robust. The latter would be more appropriate for a company managing a large suite of businesses and products.

Integration

This pertains to the need for existing systems to interact with the new software, and how well the new software allows it. Companies looking to revamp their systems entirely will be more willing to accept a solution that locks them into a particular platform versus those that use an array of software that runs before, next to, or after the projection system. The latter type of company should favour software that plays nicely with others.

So, how does one choose the best cash flow modelling system? Like so many things in life (insurance), it depends on the insurer’s specifics and the objectives of the modernisation exercise.

The actuaries at Insight Life Solutions are conversant in a number of the most popular actuarial cash flow projection software platforms and, given our additional IT and computer science qualifications, are well placed not only to help select the best solutions for clients but also to support migration and implementation projects.  Email Head of Insight Life Solutions, Pamela (pamelah@insight.co.za) or our Head of Actuarial Systems, Garrit (garritn@insight.co.za), to discuss the next steps in your modernisation journey.

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)