Insurance risk regulation has quietly changed shape. The frameworks that rating agencies and regulators apply to insurers no longer accept a simplified estimate of what a portfolio would be worth under stress. Increasingly, they expect insurers to demonstrate a genuinely recalculated value — every instrument, repriced under each prescribed condition.
Two frameworks make this especially clear: AM Best’s and the New York 7 interest-rate scenarios under Regulation 126. Understanding what each one asks for shows why the old shortcut — approximating a shock with duration and convexity — is running out of road, particularly for mid-market insurers without large in-house quant teams.
BCAR — Best’s Capital Adequacy Ratio — is the quantitative core of how AM Best assesses an insurer’s balance-sheet strength. Rather than a single number, BCAR is calculated at several value-at-risk (VaR) confidence levels (95%, 99%, 99.5%, and 99.6%, with a 99.8% level used to discuss tail risk in the ERM review)reaching well into the tail of the distribution, so that an insurer’s capital position is tested against increasingly severe outcomes.
Those outcomes aren’t guessed. Those factors aren’t arbitrary. AM Best calibrated BCAR’s required-capital factors using stochastic modeling of interest rates, equity returns, bond defaults, and real-estate movements, so the charge at each confidence level reflects an increasingly severe tail outcome. In practice, an insurer’s BCAR score is produced by applying those calibrated factors (which for fixed income vary by rating and maturity) to a portfolio marked to market. Alongside the score, the rating process tests a standard BCAR against a stressed BCAR and takes a forward-looking ‘as-will-be’ view, and this is where company-specific scenario analysis comes in. AM Best also compares a standard BCAR against a stressed BCAR, and evaluates insurers on both an “as-is” and a forward-looking “as-will-be” basis.
What this asks of the asset book: to produce a defensible market-value distribution deep in the tail, an insurer has to produce defensible marked-to-market values — and, for the stressed and ‘as-will-be’ views, a well-supported picture of how those values move under specific scenarios, rather than extrapolating one estimate from a starting point. The further into the tail the assessment goes, the more a linear approximation drifts from the true repriced value.
The “New York 7” refers to the prescribed interest-rate scenarios that New York’s Regulation 126 requires as part of life insurers’ asset-adequacy analysis. The exercise tests whether the assets supporting a block of business remain adequate under a defined set of interest-rate paths.
Those paths come in two shapes, and both matter:
The current cycle also layers in additional sensitivity testing, including a large pop-up scenario, and requires insurers to report the present value of ending surplus for each scenario.
What this asks of the asset book: the multi-year paths force each instrument to be “life-cycled” forward — its cash flows projected year by year and its value recomputed at each step along the rate path. A single duration bump cannot represent a path that bends over ten years; the position has to be repriced as it evolves.
Strip away the differences, and AM Best and the New York 7 ask for the same underlying capability from an insurer’s investment portfolio: the ability to fully reprice every instrument under a defined stress, rather than approximate the effect, — producing the repriced asset values and cash flows the analysis is built on.
This is where the traditional approach shows its age. Estimating a shock from duration and convexity is reasonable when the move is small — 10 or 20 basis points. But the shocks these frameworks specify are not small. Prescribed regulatory moves of 100, 300, or more basis points, and capital assessments that reach deep into the VaR tail, sit far outside the range where a linear estimate stays accurate. At that magnitude, an approximation is just that — an estimate that widens as the shock grows, and one that is difficult to defend when a regulator or rating analyst asks how a specific stressed value was produced.
The problem is compounded by what insurers now hold. Portfolios have moved well beyond plain-vanilla bonds into structured products, derivatives, and other complex instruments whose behavior under stress is non-linear by nature. These are precisely the positions where duration-based approximation breaks down — and precisely where full repricing is required to get the answer right.
Full repricing means computing the value of each position from its own terms and current market conditions, then changing those conditions and rebuilding the value — “bump and reprice.” Shock the interest-rate curve, and the underlying observable instruments are shocked, the curve is recalibrated, and every position is revalued against it, from the bottom up.
Because every stressed value traces back to its inputs, the output is inspectable — the honest answer to “show me how you calculated that.”
The hardest position to be in is facing large-carrier expectations with mid-market resources.
A mid-sized insurer answers to the same regulators and rating agencies as the largest carriers, and increasingly holds the same portfolio complexity — but rarely has the same quant infrastructure to build and maintain a full repricing engine. Which is why so many still rely on on-premise, spreadsheet-adjacent tooling and duration approximations that the frameworks are steadily outgrowing.
That gap — between what the frameworks now expect and what mid-market tooling can produce — is the challenge worth watching over the next few years. The insurers that close it will be the ones who can answer a regulator’s “show me how” without a fire drill.
Clearwater works with insurers on investment data, accounting, and regulatory reporting. Learn more about our insurance solutions.