The Structure of Shared Risk and Its Data
In one line
Insurers take out insurance too. That contract is reinsurance, and the amount the primary insurer actually bears comes out only after subtracting the cession, and if the cession data does not match the ledger, there is that much of a hole in the financial statements.
Why this was needed
Insurers sell contracts where a single accident can cost billions of won. If several such cases cluster in a year, the company is shaken. So insurers pass part of their own risk to other insurers. This is reinsurance.
Let us sort out the terms first.
원수사 고객과 직접 계약한 회사
재보험사 원수사의 위험을 받아 주는 회사
출재 원수사가 위험을 넘기는 것 (내보낼 出)
수재 재보험사가 위험을 받는 것 (받을 受)
In the field, "cession data" means the details of the risk our company passed to the outside.
How it works
Reinsurance splits broadly into two.
비례(proportional) 손해액의 정해진 비율을 나눈다
Quota Share (QS) 모든 계약에서 같은 비율. QS30 이면 30% 출재
Surplus 보유 한도를 넘는 부분만 비례로
비비례(non-proportional) 손해액의 크기에 따라 나눈다
초과손해액 (XOL) 사고당 일정 금액을 넘는 부분만 재보험사가 부담
Stop Loss 연간 손해율이 일정 수준을 넘는 부분을 부담
This difference shows up in large accidents. Proportional reinsurance takes off the same ratio regardless of size. Even if a 1.2 billion won accident occurs, with QS30 only 360 million won passes over and 840 million won stays as our share. That is how a single large case comes to account for 24% of the half-year net retained loss. To prevent such cases, you need an excess-of-loss treaty that puts a per-accident limit.
And there is a problem on the data side. The cession details usually exist in two places.
재보험 관리 시스템 건별 출재 내역. 사람이 등록하는 경우가 많다
회계 원장 월 단위 출재 채권 계상액. 원수 데이터에서 계산된다
The two do not match. The cession filing itself is omitted, the cession rate is entered wrongly, or a limit is applied wrongly to a large case and the amount goes off. The difference is that much of a hole in the financial statements.
Reconciliation is done in two steps. First you narrow down to the accident-month level to find the months whose difference is not 0, and then you open that month at the per-case level to pin down which case it diverged on. If you stop at the month, all you can say is "September does not match," and with that, nobody can touch anything.
There is one judgment error that comes up often here. When most of the difference amount comes from a single case, there is a temptation to skip over the small remaining ones. But the small-amount cases usually come from the same cause (manual entry), and that cause repeats as it is next month. A large case ends once fixed, but the small ones end only when the structure is fixed.
What it looks like in the field
First, there are many clients whose loss ratio reports mix the gross basis and the net retained basis. In the same meeting, one person speaks of the gross loss ratio and another of the net retained loss ratio, and both just call it "the loss ratio." The numbers differ by 20–30%, and nobody finds it odd. Making each report state its basis is the first recommendation.
Second, when a large accident occurs, it is common for the cession handling of that one case to be delayed. It is because the amount is large and the conditions are complicated, so negotiation with the reinsurer is attached. If the close passes in the meantime, the net retained is booked larger than it really is and comes back the next quarter.
Third, I often see cases where the cession reconciliation is off every month and left alone for years because the amount is small. Then when a single large case goes wrong through the same path, it is in the hundreds of millions. If the structure is the same, only the size differs, and it is the same incident.
What you will do in the next lab
You reconcile the accounting ledger and the reinsurance system on 1,296 losses covered by QS30 proportional reinsurance. You narrow the difference of 263,630,000 won to five accident months and then pin it down to five cases, of which 98.6% comes from one large case. Then you compute the share the large case takes in net retained loss and see in numbers why proportional reinsurance alone is not enough.