standard formula vs internal model

Imagine two ways to weigh a parcel. One is a public coin-operated scale at the post office: anyone can use it, it is calibrated the same for everybody, and it is roughly right for typical packages. The other is your own precision balance, tuned to your exact products — more accurate for what you actually ship, but you had to build it, prove it works, and let an inspector certify it. Choosing how to compute the Solvency Capital Requirement is exactly this choice: the standard formula (the public scale) versus an internal model (your own certified balance).

The standard formula is the regulator's prescribed recipe for the SCR: a fixed menu of risk modules (market, life, non-life, health, counterparty default, operational) each computed with given stresses and parameters, then combined with prescribed correlation matrices that grant a set diversification benefit. It is calibrated for an 'average' European insurer and is simpler, comparable, and cheaper to run. An internal model is a firm's own bespoke model of its risks — often a full stochastic simulation — which must be approved by the supervisor and pass strict tests, above all the 'use test': proof that the model is genuinely used to run the business, not merely to lower the capital number. Firms can also use a partial internal model, mixing the two: bespoke where they differ from the average, standard elsewhere.

Why it matters: the standard formula can over- or under-state capital for an insurer whose risk profile departs from the average — a monoline catastrophe writer, or a firm with unusual guarantees — so a well-built internal model can produce a fairer (and sometimes lower) SCR while sharpening management's own understanding. The honest caveats are real: internal models are costly to build, validate and maintain; their approval is demanding; and there is an ever-present temptation to tune assumptions toward a flattering capital figure, which is exactly why the use test and ongoing supervisory scrutiny exist.