Link to the search page

Introduction to Bermuda SBA Modeling - Part 2

In this article, we continue to discuss the modeling considerations for Bermuda SBA BEL calculations.

et-2025-06-yuan-hero.jpg

In the first installment of our two-part series, we provided an overview of Bermuda’s economic balance sheet (EBS) and covered three modeling considerations for scenario-based approach (SBA) modeling: Modeling assets and reinvestment, stress testing and solving for the starting asset market value. In this installment, we dive into the remaining three modeling considerations.

1. Model Validation

As part of consultation paper 2 (CP2), the Bermuda Monetary Authority (BMA) requires validation to be performed independently (externally or internally) for the SBA model. This section describes some testing approaches for an SBA model. This is akin to a “second line” validation concept as part of a model risk management framework, which provides effective challenges and independent review.

Scope of Validation

After the SBA model is set up, one should consider testing the iterative solving mechanism under different market environments (e.g., low interest rates, extreme markets, company’s own ERM stresses). The BMA has listed 13 items that could be checked during the model validation, such as trend analysis, controls verification and cell testing.

It is also important to test asset cash flows under different scenarios. The SBA model should capture the optionality of the assets and model features of assets appropriately. One should also check that the closing market value of assets at the end of the projection is immaterial.

A critical aspect of validation is determining the scope and coverage of the exercise, as outlined in CP2. This includes defining the specific components of the SBA model to be validated, whether it’s the data inputs, assumptions, mathematical algorithms, or the model’s overall performance in various scenarios.

BMA Requirement

When selecting a validation team, whether internal or external, it’s essential to review that its members have the necessary independence, the relevant expertise in SBA calculation and Bermuda regime, and are equipped with the skill sets to provide an effective challenge.

The BMA’s requirements also encompass governance and controls. Companies must establish robust control mechanisms to monitor and manage the SBA model continually, alongside any upstream and downstream models. This includes model risk management policies, model change policies and environments, procedure documentation, and review and oversight structures.

2. Forecasting

Forecasting within the context of a Bermuda SBA model presents a set of unique challenges. While firms may want to forecast the SBA best-estimate liability (BEL) for several uses such as financial planning, pricing, risk margin and stress testing, the modeling process is inherently complex. It potentially involves nested projections, whereby assets and liabilities are reprojected along with the SBA calculation at each projection timestep. This allows for changing economic environments as well as experience emergence on the liabilities in consideration.

Runtime Considerations

One of the primary challenges in forecasting is the runtime for large in-force blocks. This necessitates a careful balance between precision and efficiency, as excessively long runtimes can be impractical for real-time decision-making and strategic planning.

When considering a nested projection, the model needs to solve for starting assets for each inner calculation, capture asset interaction in nested calculations, and reflect asset and liability interaction. This run can be difficult, even leveraging linear computations (i.e., distributing calculations across multiple computers), and will be time-consuming. Additionally, running lapse shocks in the projection for Bermuda solvency capital requirements (BSCR) presents further run requirements. All the modeling issues mentioned previously would be exacerbated in a forecasting setting.

A simplified approach to forecasting reserves is to discount liability cash flows based on the spread above risk-free yields, solved during the initial SBA reserve calculations. Any expected changes in the risk-free rate and spread can be reflected within the forecasting of the economics to project the BEL. This approach can result in projected reserves that mirror closely to a first-principles SBA projected reserve, but this method should be closely evaluated where this approximation may no longer be appropriate (e.g., as asset investment strategy shifts/changes over time, optionality in assets and liabilities that would require reprojection rather than assuming the liability cash flows are static).

Assumption Shocks

Application of assumption shocks present additional forecasting challenges. For instance, modeling lapse shocks for both lapse risk BSCR and the new lapse cost (LapC) introduced at multiple time points are a crucial consideration. These shocks should be captured accurately to reflect the potential variations in policyholder behavior, thereby affecting the financial outcomes of the SBA model.

Risk Margin

Furthermore, the risk margin portion of the technical provisions requires forecasting the BSCR, which adds another layer of complexity. Here, modelers face the challenge of approximating or utilizing nested inner-loop projections to ensure that capital charges are reliably estimated. An example is the projection of BSCR longevity risk, which increases over time as the expected attained ages of policyholders increases, potentially requiring seriatim level liability and BEL projections.

3. Integration with Existing Models

When implementing an SBA calculation, another key consideration is how it integrates with existing models. The SBA model may be in a separate platform, model, or spreadsheet tool, especially if a “minimum viable product” modeling approach was taken to gauge financial impact or assess pricing at the forefront of prospective new deals. Companies in these situations will need to decide whether they want to keep the model separate or integrate with more established production models.

The SBA can be a useful tool when integrated into the decision-making and management of the business, rather than being viewed as a model to satisfy regulatory requirements. The model becomes an even more powerful tool if integrated with the asset and liability models, as this could allow firms to perform more holistic stress testing from just one place (rather than using asset, liability and then SBA calculations separately).

Operational Considerations

The model change management considerations also apply as they do to any actuarial model producing reserve calculations. Careful planning, model merges, and regression testing need to be performed such that the integrated model can serve multiple purposes for the US and Bermuda. It is crucial to maintain consistency and reliability in the integrated model.

In addition to the technical aspects, operational considerations come into play, such as model ownership. Decisions need to be made regarding who will be responsible for the SBA model and its maintenance, as well as how it will be used within the company’s broader operational framework. Clarity in ownership and operational procedures is vital for a seamless transition and ongoing management of the SBA model in the Bermuda market.

Conclusion

The consultation papers bring forth a distinctive set of challenges and considerations for SBA model development and associated validation. These considerations encompass a spectrum of intricate factors, from the solving for starting asset market values, use of enhanced reinvestment/disinvestment strategies, the need for independent validation, and complex forecasting techniques. Insurance companies operating in Bermuda must navigate these intricacies with precision and foresight to ensure regulatory compliance, financial soundness, and risk management that aligns with the specific demands of this vibrant insurance market.

The views reflected in this article are the views of the authors and do not necessarily reflect the views of Ernst & Young LLP or other members of the global EY organization. Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries, the newsletter editors, or the respective authors’ employers.