The number you did not get to choose, and the one you do
Everything you built in the previous four guides of this rung pointed at a single, hard, externally imposed number. You met ERM as the firm-wide view, learned that a value at risk marks a cliff edge while tail VaR tells you how far you fall beyond it, turned a tolerable bad outcome into economic capital, and saw the Solvency II machinery grind that into the regulator's official Solvency Capital Requirement. All of that is the regulator asking *one* standardised question: under this prescribed 1-in-200 shock, can you still pay? It is necessary. It is also not enough.
The trouble is that a standardised formula, by design, does not know *your* firm. It does not know that your book is two-thirds catastrophe-exposed coastal property, that your largest reinsurer is also your parent, that a third of your profit hangs on a single guaranteed-rate product, or that your growth plan triples your longevity exposure over three years. The Own Risk and Solvency Assessment — the ORSA — is the place where the firm finally gets to answer in its own voice. It is the one number, or rather the one *process*, you do get to choose: the company's own forward-looking view of all its material risks and the capital it judges it needs to carry them.
Forward-looking: rolling the balance sheet through time
What does "forward-looking" actually mean in practice? You take the business plan the board has approved — premium growth, new products, the planned dividend, the investment strategy — and you *project the whole balance sheet forward* year by year across the planning horizon. At each future year-end you re-run the question you already know how to ask: are the projected assets still enough to cover the projected liabilities plus the capital requirement that those future, larger liabilities will themselves demand? The ORSA is, at heart, the actuarial control cycle applied to the entire firm: project, compare against the plan, and feed what you learn back into the plan.
A tiny picture makes it concrete. Say today you hold own funds of 150 against an SCR of 100 — a comfortable solvency ratio of 150%. But your plan writes a lot of new business, and new business is capital-hungry: each year's growth pushes next year's SCR up faster than the retained profit refills the funds. Project it forward and you might find the ratio drifting 150% → 138% → 129% → 121%. No single year breaches the requirement, yet the *trajectory* is walking steadily toward the edge. A point-in-time SCR can never reveal that slope; only the forward roll can. That slope is exactly the kind of thing the ORSA exists to surface — early, while there is still time to act.
Forward-looking solvency projection (own funds / SCR):
Year-end Own funds SCR Ratio Note
-------- --------- ---- ------ ----------------------------
Now 150 100 150% comfortable
+1 152 110 138% growth lifts SCR faster
+2 155 120 129% than profit refills funds
+3 157 130 121% no breach... but the
*trajectory* heads downhill
The SCR is a photo of 'Now'. The ORSA is the whole column.Stress and scenario testing: breaking the plan on purpose
A forward projection of the *expected* plan is comforting and slightly useless on its own — the future will not be the central estimate. So the engine of the ORSA is stress and scenario testing: you deliberately break the plan to see whether the firm survives the breaking. A stress test twists one assumption hard — equities fall 40%, interest rates drop 100 basis points, mortality worsens 20%, the loss ratio jumps 15 points — and you re-roll the projection to watch what the solvency ratio does. A scenario test is richer: it builds an internally coherent *story* in which several things go wrong together, the way they actually do in a real crisis.
The distinction matters more than it looks, because real ruin almost never comes from one variable moving in isolation. In a true downturn, equities fall *and* credit spreads widen *and* lapses spike *and* the reinsurer you were counting on is itself under strain — and the correlations you assumed in calm times quietly climb toward one. A scenario captures that conspiracy; a one-at-a-time stress, summed up, badly understates it. This is why the SCR's neat diversification benefit — the capital relief you get because risks are imperfectly correlated — is the first thing a good ORSA puts under suspicion: diversification is exactly what evaporates in the crisis you are stress-testing for.
The use test: a model that does not change anything is just decoration
Now we reach the rule that holds the entire structure honest, and it is deceptively short. The use test says: the firm's risk model — its internal model, its capital number, its ORSA — must *genuinely be used to run the business*, not merely produced once a year to satisfy the regulator and then shelved. If a firm wants supervisory approval to replace the standard formula with its own internal model (almost always because the internal model gives a lower, better-tailored capital figure), it must *earn* that privilege by proving the model is woven into real decisions. You cannot have the lower number without living by the model that produced it.
What does "actually used" look like? It is concrete and auditable. The model's view of how much capital a line of business consumes feeds the *pricing* — a capital-hungry product must earn back its cost of capital in the premium, or you should not write it. It shapes *reinsurance buying*: the layers you cede are the ones the model says are eating the most tail capital. It informs the *investment strategy*, the *dividend* the board declares, the *limits* set under the firm's risk appetite, and which lines the firm grows or shrinks. When a board paper proposes a new product, the model's capital and return numbers should be *on the page* — and should sometimes change the decision. A model that has never once said "no, don't do that" is not being used.
There is a deep reason the regulator insists on this, beyond bureaucratic tidiness. A model nobody bets real money on never gets stress-tested by reality, so its errors fester unseen; a model that drives pricing and reinsurance gets challenged the moment its numbers feel wrong to someone whose bonus depends on them. The use test harnesses self-interest to keep the model honest. It is also what finally makes the actuary's work matter: it forces the chain you have climbed this whole ladder to build — from a single ruin probability, through capital, to the boardroom — to actually close. The model only earns its keep when it changes what the firm does.
Honest about the limits — and where the ladder leaves you
Carry the same humility into the ORSA that you carried into every model on this ladder. Stress tests can only break the firm in ways someone *thought to imagine*; the scenario that actually sinks a company is, with grim regularity, the one nobody wrote down — the unknown unknown. A scenario library curated only from past crises is a museum of the last war, not a map of the next. And reverse stress testing, powerful as it is, can lull you into believing that because you found the failure points you can imagine, you have found them all. You have not. The danger is precisely the failure mode that never made it onto the page.
There is also a quieter institutional failure mode the use test is meant to fight but cannot always defeat: the ORSA that becomes a beautifully bound document produced to be filed, not to be read. When the projections are reverse-engineered to land on a comfortable answer, when the scenarios are chosen to be survivable rather than searching, when the board nods the report through unread — then every box is ticked and nothing has been learned. The discipline lives or dies on *intellectual honesty*, which is why it sits so close to the profession's code of conduct and the actuary's duty to speak uncomfortable truths to people who would rather not hear them.
And so the whole ladder closes here. You began with a single uncertain claim and a coin-flip's worth of probability; you learned to discount money through time, to read a life table, to price a policy, to set a reserve, to fit a loss distribution, to measure ruin, and finally to gather every one of those risks onto a single balance sheet and ask what capital must stand behind the firm's promises. The ORSA is where all of it converges into one living judgement, renewed every year, owned by the board, and tested against a future nobody can see. The mathematics gave you the answer; the use test makes you act on it; and your honesty about its limits is, in the end, the most actuarial thing about you.