Parts 1–3 have exact answers. Parts 4 and 5 do not — compare whether yours answers the same questions this one does.
1 · Before and after
today: 890 / 555 = 160.4% 🟢
after 300m: 590 / 555 = 106.3% 🟢 — still green, but the headroom is nearly gone
A dividend reduces risk-bearing capital and leaves target capital untouched: the book has not changed, only the capital behind it.
2 · The largest green dividend
Green requires the ratio to exceed 100%, so solve:
(890 − D) / 555 > 1 ⟺ D < 890 − 555 = 335
The largest dividend to the nearest million that stays green is CHF 334m — at exactly 335 the ratio is 100.0%, which Art. 51 §2(a) puts in yellow, not green (green is above the threshold).
3 · Does it need approval?
Both do. Art. 52 §1 requires prior FINMA approval for solvency-relevant actions if the company would not be in the green band immediately after them — and a dividend is named explicitly, alongside capital repayments, commuting outward reinsurance and group-internal transactions.
- The proposed 300m leaves the company at 106.3%, i.e. in green, so Art. 52 §1 does not bite on that ground.
- The 334m maximum also leaves it green (just), so the same holds.
That is the letter of it. In practice a company proposing to spend 99% of its headroom is having a conversation with its supervisor whether or not a form requires it — and if any input turns out to have been optimistic, the action that looked green was not.
4 · Current position or projected position?
For projecting:
- The dividend is paid in nine months, and the test that matters is solvency at the moment capital leaves. A reference-date ratio is nine months stale by then.
- The company will have earned premium, paid claims and possibly moved its reserves in between. Ignoring the whole year's expected result — which the SST itself explicitly recognises, see Lesson 2 — tests a position that will not exist.
Against relying on a projection:
- A projection is a forecast made by the party that wants the answer to be yes. The reference-date position is audited, filed and adversarially reviewed; the projection is neither. Where the two disagree, the incentive is not symmetric.
The practical resolution is the one the ordinance itself takes: file on the reference date, and Art. 50 §3 requires immediate notification if the risk situation changes enough to reduce the ratio materially — including falling below an intervention threshold — with an approximate recomputation attached. The projection informs the decision; it does not replace the test.
5 · Judgement — one defensible answer
I would advise against 300m, and propose something closer to 150–200m, for three reasons.
- The headroom is the shock absorber, and this book has a long tail. Motor liability develops over decades and is the classic carrier of the two risks that move an SST ratio without anyone underwriting anything: claims inflation and a change in the discount rate. A company at 106% has almost no room for either.
- The ratio is a point estimate of a modelled quantity. Target capital is itself the output of a model with selected parameters — Lesson 4 will show that the single most influential number in the reserve component is a coefficient of variation the actuary chose. Treating 106% as if it were measured to three digits is a category error.
- The asymmetry of being wrong. Paying 150m and being able to pay more later costs the shareholder a delay. Paying 300m and finding the ratio at 90% costs a measures plan, a supervisory conversation, and the loss of the freedom to act without approval.
What I would want to see first:
- The ratio's sensitivity to the assumptions that actually drive it — at minimum a claims-inflation shock and an interest-rate shock, and the reserve-risk coefficient of variation moved to its benchmark.
- The projected position at the intended payment date, with the expected insurance result and the planned new business in it.
- A statement of what the company would do if the ratio landed at 90% — because if there is no credible answer, the headroom was not spare.