How insurers can improve fairness, efficiency and control through automated bonus rate setting.
Bonus rate setting is one of the most critical yet time-consuming tasks for life insurers. Traditionally reliant on expert judgment and manual iteration, it consumes valuable actuarial time and introduces inconsistency. By applying automation, this process can be transformed, delivering faster, fairer and more transparent outcomes.
Actuarial and financial models are often highly complex, producing outputs calculated to the tenth decimal place and projecting results nearly a century into the future. Yet, despite this sophistication, some critical decisions, such as determining appropriate bonus rates, still rely heavily on manual judgment. This process is typically iterative, resource-intensive and time-consuming.
In an era where artificial intelligence and automation are reshaping expectations of efficiency, there is an opportunity to enhance existing models to calculate bonus rates automatically. Drawing directly on the existing inputs and calculations in these models to set bonus rates not only reduces manual effort but also unlocks significant time and resources for higher-value analysis and decision-making.
Where Bonus Rates Are Used
Understanding where bonus rates fit within the wider insurance framework is essential before exploring how automation can improve the process.
Bonus rates play a central role in the products offered by mutual insurers and other participating life funds. In these structures, policyholders are not only customers but also share in the profits generated by the insurer. Instead of distributing all earnings directly as cash, insurers often declare bonuses that are added to the policy benefits.
These bonuses serve several purposes:
- Sharing profits fairly – ensuring policyholders benefit from the performance of the fund.
- Smoothing outcomes – using discretion to stabilise returns over time, avoiding sharp swings that might occur if profits were passed on directly each year.
- Strengthening engagement – reinforcing the value proposition of participating policies by demonstrating tangible returns.
Since bonuses directly affect both policyholder value and the insurer’s financial stability, setting the appropriate rates is a critical decision. It requires balancing fairness to policyholders, prudence for the insurer and consistency with long-term objectives, which is why it has traditionally been left to expert judgment.
How an Automated Bonus Solver Works
At its core, the automated bonus solver replaces manual judgment with a structured algorithm that finds the right rate automatically. An automated bonus solver allows insurers to formalise and streamline the process of setting bonus rates. The user first defines the key parameters and the target outcome that the bonus rate should achieve, typically a well-defined solvency threshold.
The solver then applies an iterative process:
- Initial projection using a user-defined bonus rate – In the first iteration, the assets and liabilities are calculated using a user-defined bonus rate, typically the previous year’s declared bonus rate. This provides a starting point for the solver to assess the financial position under current conditions.
- Check sufficiency – The solver then tests whether the available funds are adequate to meet the defined solvency target.
- Borrowing logic – If permissible, the solver can draw on other available funds to close any shortfall.
- Estimate revised bonus rate – Finally, the solver uses interpolation techniques to estimate a bonus rate that would bring the financial position into balance. This estimate is based on the relationship between the current financial position (from the sufficiency check) and the current bonus rate used in the liabilities calculation.
Once a revised bonus rate has been estimated, the solver re-runs the projection, checks for sufficiency and repeats the process. This continues until the difference between the required and actual outcomes is within an acceptable tolerance, ensuring convergence to the correct bonus rate.
This transforms what was once a manual, trial-and-error exercise into a systematic, repeatable and fully traceable process embedded directly in the model.
When the relationship between the bonus rate and the financial position is approximately linear, the solver will typically converge within 2-3 iterations, as the interpolation step provides an accurate estimate of the required adjustment. In cases where the relationship is non-linear, more advanced mathematical techniques are applied to accelerate convergence. Algorithms such as the Newton–Raphson method, which use first and second derivatives to refine estimates, are well suited to this problem, allowing the solver to reach the correct bonus rate efficiently, even in complex, non-linear settings.
Dynamic Projections
While the solver described above focuses on determining an appropriate bonus rate at a single valuation date, actuarial teams are often equally interested in how these rates may evolve over time. This longer-term view is referred to as the bonus path, the projected trajectory of future bonus rates that reflects expected experience on the insured portfolio and allows actuaries to assess the financial dynamics under a range of scenarios.
In many actuarial models, this is achieved through dynamic projections, where the bonus solver is applied sequentially at each future time step across the projection horizon. The process begins at the valuation date, where the model solves for the bonus rate required to meet defined conditions e.g. achieving target policyholder returns or maintaining solvency coverage. Once this rate is determined, the model rolls forward the assets and liabilities by one period, and the solver is applied again at the next time step. This cycle continues for each year of the projection horizon, which can extend to 100 years or more.
A critical consideration is that each forward step requires a full prospective valuation of liabilities. For a 100-year projection:
- The first valuation (at time 0) projects cash flows over 100 years.
- The second valuation (at time 1) projects 99 years.
- The third (at time 2) projects 98 years.
- …and so on, until the final step which involves only one year.
The total computational effort is therefore:
100 + 99 + 98 + … + 1 = 5,050
In other words, a 100-year projection involves 5,050 point-in-time calculations. When multiple scenarios or sensitivities are required, for example, under different economic or demographic assumptions, the computational load multiplies rapidly, creating a significant performance bottleneck.
Solving the Time Vector at Valuation
To overcome the computational burden of dynamic projections, an alternative approach i.e. the time vector method, replaces the sequential year-by-year process with a single, multi-period solution determined at the valuation date.
Instead of solving for bonus rates one year at a time, the actuary defines the entire set of future bonus rates as a time vector, which represents the expected trajectory of bonus rates over the full projection horizon. This vector is then determined through an iterative process carried out entirely at the valuation date.
The process begins with an initial guess for the future bonus rates. Assets and liabilities are then projected once over the full 100-year horizon, and the results are tested against the required conditions at each future time step e.g. solvency targets or policyholder return thresholds. The bonus rate vector is subsequently adjusted, and the process is repeated until the conditions are simultaneously satisfied across all future years.
Since each iteration involves only one full 100-year projection, the total computational effort is dramatically reduced. Assuming convergence within five iterations, typical in practice, the process involves approximately 500 calculations, compared to the 5,050 required under the dynamic projection approach.
This represents an order-of-magnitude improvement in efficiency while preserving accuracy and internal consistency. The time vector method also simplifies scenario testing and reduces operational complexity, as the entire bonus path is determined upfront rather than through hundreds of interdependent projection loops.
Importantly, while the dynamic projection functionality requires additional and often costly licences from actuarial software vendors to handle iterative projections, the time vector approach can typically be implemented using standard functionality within existing actuarial modelling platforms. This means it can sit alongside existing model run setups and operational processes without requiring extensive system changes or additional software expenditure, making it both a technically and commercially attractive solution.
Benefits of Using an Automated Bonus Solver
Automating bonus rate setting doesn’t just streamline actuarial work — it drives real business outcomes.
Although organisations often expect the development and implementation of an automated bonus solver to be costly and complex, experience has shown that the benefits typically outweigh the investment by a significant margin. Once embedded, the solver delivers substantial time and resource savings in each valuation cycle. As a result, the payback period is usually between 6 and 18 months, depending on the complexity of the bonus algorithm and the projection framework. In addition, the computational savings realised by implementing an automated bonus solver deliver tangible benefits not only to the actuarial function but also across the insurance firm.
Key benefits include:
- Improved accuracy and consistency – Since bonus rates are derived through a systematic, algorithmic process rather than repeated manual calculations, results are both more accurate and internally consistent. The time vector method also ensures that the entire future path of bonus rates evolves coherently across the projection horizon.
- Reduced operational risk – Automation removes the dependency on repeated manual judgment at each time step. The process becomes fully governed, reproducible and auditable.
- Scalability and flexibility – Whether applied to different funds, products or geographies, the automated solver can be scaled and adapted without exponentially increasing computational effort.
- Faster decision-making and scenario analysis – The significant reduction in model run-times enables actuaries to run multiple scenarios, sensitivities or stress tests with ease, supporting better strategic decision-making.
- Future-proofing actuarial models – By embedding bonus rate setting logic directly into existing models, insurers lay the groundwork for more advanced automation and AI-driven techniques in the future, enhancing both speed and strategic agility.
Conclusion
Despite the sophistication of actuarial and financial models, critical decisions like setting bonus rates have long relied on manual judgment and iterative processes. By embedding an automated bonus solver into existing models, insurers can transform this task from a time-consuming exercise into a streamlined, repeatable and transparent calculation.
Moreover, extending this approach to solve entire bonus paths as a time vector offers a powerful enhancement over traditional dynamic projections. It dramatically reduces the number of runs required, improves computational efficiency and provides actuaries with a robust framework to assess future bonus rate evolution under a range of scenarios.
The result is not only significant efficiency gains and reduced operational risk but also the opportunity for actuarial teams to shift their focus from routine tasks to higher-value activities that drive long-term strategy. In an industry where both efficiency and fairness are paramount, automation offers a practical way forward. This enables insurers to deliver consistent outcomes for policyholders while optimising the use of scarce actuarial resources.
Automation is redefining what’s possible in actuarial modelling. Firms that embed these techniques now not only gain efficiency today but also build the foundation for next-generation AI-enabled decision-making.
To explore how MBE Consulting can help integrate automated bonus solvers or broader model automation into your actuarial framework, contact our team.
