Navigating the Complexities of VM-22: How Standardized Software Solves the Computational Challenge of Non-Variable Annuity Reserving

Authored by John Bowers, Head of Actuarial Product and Consulting at RNA Analytics

The implementation of Principle-Based Reserving (PBR) under Valuation Manual 22 (VM-22) represents a transformative shift for life insurers managing non-variable annuity products, including Fixed Annuities, Fixed Indexed Annuities, and Payout Annuities. Moving away from formulaic standard approaches toward dynamic, principle-based asset-liability stochastic valuations presents a profound operational and computational hurdle for actuarial teams. Life insurers are tasked with projecting both liability and supporting asset cash flows across hundreds of stochastic scenarios while navigating multiple regulatory valuation bases simultaneously. As reporting deadlines tighten, the primary pain point for insurers is no longer just understanding regulatory compliance it is building an agile, highly scalable actuarial modeling infrastructure capable of processing massive data sets efficiently without crashing workflows or ballooning runtime overheads.

At the heart of the operational strain caused by VM-22 is the requirement to run heavy stochastic simulations record-by-record across distinct calculation demographic assumptions. Actuaries must account for diverse perspectives: the ’Best Efforts’ basis reflecting company future hedging strategies; the ’Adjusted’ basis utilizing prudent estimate assumptions; and the ’CTEPA’ basis executing prescribed standard projections with regulatory parameters. Furthermore, additional calculations and disclosure for riders including the Guaranteed Actuarial Present Value (GAPV) andVM-31 Additional Standard Projection Amount (ASPA).  Dynamic index-crediting mechanics adds further complexity. Managing these multifaceted layers traditionally meant dealing with unwieldy actuarial models and computational bottlenecks that forced teams into lengthy overnight or multi-day runs.

To solve these core operational pain points, software architecture must evolve alongside regulatory mandates. Through the R3S US Standard Code VM-22 framework, RNA Analytics directly addresses these computation and scalability challenges with a highly flexible, cloud-agnostic architecture built to scale effortlessly across any cloud environment or multi-cloud ecosystem. Before outputting to storage, rather than generating heavy, file-based workspace outputs that clog local environments and cloud networks, the R3S core architecture writes scenario-specific cash flows directly to external databases via optimized connection strings. This design dramatically improves execution performance, allowing system runtimes to scale below direct record multiples as policy data volumes grow. In benchmark runs across 100 scenarios and three valuation layers, hardware configurations ranging from cloud instances like Azure 2 (16 CPU, 64GB RAM) down to standard laptops (Ultra 7 155U, 32GB RAM) demonstrate sub-linear scaling, processing 10,000 to 100,000 policy records in manageable timeframes.  

Beyond architecture-level performance, flexibility in calculation methodology is crucial for insurers optimizing their capital and reserve levels. The R3S VM-22 framework provides robust support for both standard and advanced valuation approaches. Insurers can leverage the Net Asset Earned Rate (NAER) Method to project liability cash flows and establish accumulated deficiencies per scenario. Alternatively, teams can utilize the Direct Iteration Method (DIM), which employs specialized R3S targeting algorithms to calculate the exact starting assets required to cover liabilities, given a specific portfolio of investment assets with zero deficiencies. By integrating comprehensive option earned income modeling and dynamic cap rates for Fixed Indexed Annuities into a unified Standard Code model, actuaries can move away from fragmented legacy workarounds. Ultimately, adopting purpose-built, highly scalable actuarial solutions enables insurers to turn the computational burden of VM-22 into a streamlined, repeatable process that unlocks deeper strategic insights.