學習目標 · Learning Objectives
- Explain what underwriting actually decides and who performs it — the three axes (mortality, morbidity, class/occupation), the difference between automated rules and human underwriters, and where the agent sits in that chain.
- Distinguish a pre-existing condition exclusion (PES) from a general exclusion, with Hong Kong examples, and explain why the distinction determines whether the client's most likely claim will ever be paid; then navigate the health disclosure Q&A — what "material" means, why a wrong answer invalidates the contract rather than merely raising the premium, and how to coach a client through it without coaching the answer.
- Recognise the underwriting paths (standard / simplified / full) and outcomes (standard / loaded / excluded / referred / declined), explain each outcome in commercial language the client can actually act on, and handle evidence and medical records — what to request, in what order, and how to avoid the two rework loops that blow out turnaround.
Underwriting (核保)
Need analysis (Lesson 04) tells you what the client needs. The FNA (Lesson 05) tells you what to do about it. Underwriting is where the market decides whether you are allowed to do it.
A quote assumes the client is healthy. Underwriting is the process that checks whether that assumption holds, and if it does not, decides what to do about the difference.
This is the least intuitive module in a point-of-sale system, because it is the only one where the product the agent shows the client at 3pm can be structurally different from the product the client receives at 6pm. Everything else in the POS — rates, benefits, commission — is a calculation. Underwriting is a judgement.
Learning Objectives
- Explain what underwriting actually decides and who performs it — the three axes (mortality, morbidity, class/occupation), the difference between automated rules and human underwriters, and where the agent sits in that chain.
- Distinguish a pre-existing condition exclusion (PES) from a general exclusion, with Hong Kong examples, and explain why the distinction determines whether the client's most likely claim will ever be paid; then navigate the health disclosure Q&A — what "material" means, why a wrong answer invalidates the contract rather than merely raising the premium, and how to coach a client through it without coaching the answer.
- Recognise the underwriting paths (standard / simplified / full) and outcomes (standard / loaded / excluded / referred / declined), explain each outcome in commercial language the client can actually act on, and handle evidence and medical records — what to request, in what order, and how to avoid the two rework loops that blow out turnaround.
What Underwriting Actually Decides
In practice: underwriting decides one thing — whether the carrier will take this risk, and on what terms. Everything else in this lesson is detail on that one question.
The decision is not binary and it is not about "how healthy" the client is in the abstract. It resolves into three separate axes, and a carrier may reach different conclusions on each.
| Axis | The question | Typical evidence | Who decides |
|---|---|---|---|
| Mortality | How likely is this person to die during the policy term? | Age, smoking, BMI, occupation, family history, medical history | Automated rules → referred to underwriter |
| Morbidity | If this person falls ill, how much will it cost, and how likely? | Hospitalisation history, chronic conditions, claims history | Automated rules → referred to underwriter |
| Class / occupation | Is the client's job insurable, and at what rate? | Occupation class table, employer details, duties | Automated rules → referred to underwriter |
The axes move independently. A 45-year-old non-smoker with controlled hypertension might be standard on mortality (a well-managed condition barely affects the mortality table) while being loaded on morbidity (hospitalisation frequency is genuinely higher). A commercial diver might be standard on mortality while being declined on class — the risk is not that he dies, it is that the underwriting data is not credible.
Who actually performs it
Automated rule engine Human underwriter
───────────────────────── ────────────────────
Age, smoking, BMI, sum assured, Anything the rules can't price:
occupation class, declared medical records, unusual disclosures,
conditions, routine combination risk, ambiguity,
treatments borderline sums
Speed: seconds Speed: days to weeks
Output is one of: Output is a decision note:
PASS_STANDARD "Loading 30% — hypertension
PASS_LOADED uncontrolled at declaration,
PASS_EXCLUDED FBC 6.8"
REFER_UNDERWRITER
DECLINE The note is a RATIONALE, and the
rationale is what the agent quotes
back to the client.
The agent's job at the underwriting stage is not to decide anything. It is to make the decision legible to a client who has just been told the product they agreed to buy is not the product they are getting.
That is a skill in translation, and it is the whole lesson.
The Three Words Clients Misunderstand Most
In practice: loaded, excluded, and referred sound like degrees of bad luck. They are not. They are three structurally different outcomes with three completely different consequences, and clients who conflate them will accept things they should refuse.
type UnderwritingOutcome =
| { kind: "STANDARD"; premiumMultiplier: 1; exclusions: [] }
| { kind: "LOADED"; premiumMultiplier: number; exclusions: [] }
| { kind: "EXCLUDED"; premiumMultiplier: number; exclusions: Condition[] }
| { kind: "REFERRED"; premiumMultiplier: null; exclusions: null }
| { kind: "DECLINED"; premiumMultiplier: null; exclusions: null };
// The three that get conflated:
// LOADED — "we cover you, you pay more"
// The benefit is INTACT. Every penny of it pays out.
// The client's problem is affordability over 10-30 years.
//
// EXCLUDED — "we cover you, except the thing that will most likely happen to you"
// A specific Condition[] is carved OUT of the benefit.
// The client may be paying a premium for a benefit that will
// almost certainly never be claimed.
//
// REFERRED — "we don't know yet"
// Not a mild decline. Not a soft approval. The file has left
// the automated lane and a human is now reading it.
The single most damaging sentence an agent can say to a client who has been loaded is "加費兩年後就會消失". It will not. A loading is written into the contract and appears on every future premium notice until the policy ends. Clients who believe otherwise do not cancel — they feel betrayed, and they feel betrayed for twenty years.
Pre-existing Condition Exclusion (PES) vs General Exclusion
In practice: these two look similar on a schedule and behave completely differently in a claim. Confusing them is the most common cause of a "this should have been covered" argument that no amount of documentation can fix.
A general exclusion applies to everybody. It says: this policy will never pay for this category of thing, whoever you are. Examples: war, riot, nuclear activity, pre-existing condition under a non-PES product, criminal activity, an intentional self-inflicted act.
A pre-existing condition exclusion is carved out specifically for this client, because underwriting identified a condition that was declared or detected before inception. It says: for anyone else, this condition would be covered. For you, it will not be.
| Dimension | General exclusion | Pre-existing condition exclusion |
|---|---|---|
| Who it applies to | Every policyholder | This policyholder only |
| Trigger | The nature of the event (war, riot, illegal act) | A specific declared or detected condition |
| Why it exists | Moral hazard, catastrophe exposure, insurability boundary | The risk was already present at inception; the premium priced did not fund it |
| Appears on | Standard exclusions clause, shared by all policies | The endorsement / special conditions schedule, unique to this policy |
| Removable later | Generally no | Generally no |
| Typical example | Claims arising from war or nuclear activity | IBD-related hospitalisation excluded on a medical plan |
The table above is the entire lesson. The client who asks "will you cover my [condition]?" is asking about row four versus row five. If the answer is a PES, the answer is no, permanently, and the honest thing is to say so before signature — not to let it be discovered at claim stage, eighteen months later, when the sum in question is their savings.
The asymmetry clients never see
Here is the structural problem with an exclusion, and it should be said out loud to every client who receives one:
The condition most likely to generate a claim for this client
is the one condition that has been excluded.
This is not a flaw. It is the logic of underwriting.
But it means a client with a PES is, by construction,
paying for a benefit designed around a risk that will not occur.
That sentence reframes the conversation. Without it, a client with an excluded condition hears "yes, you're covered" and stops listening. With it, they hear "you are covered for everything except the thing that will actually happen to you, and here is the price of the rest" — which is a conversation a rational client can have.
The Duty to Disclose, and Why a Wrong Answer Is Fatal
In practice: in Hong Kong insurance law, the duty to disclose is uberrimae fidei — of the utmost good faith. It is not a form-filling exercise. A material non-disclosure can allow the contract to be avoided, and the consequence is that the cover does not exist.
What "material" means
A fact is material if a reasonable insurer would have taken it into account in deciding whether to accept the risk, and on what terms. It does not need to be a death sentence for cover. A disclosed hypertension that produces a 20% loading is material. An undisclosed hypertension that produces a full rescission is worse than the loading.
The practical consequence for agents: you cannot help the client decide what is material. That is the client's legal duty, and advice on where the line sits is precisely the advice that gets an agency in trouble. What you can do — and must do — is make disclosure easy and non-threatening.
Why an incorrect answer kills the contract
Agent asks: "Have you ever been diagnosed with diabetes?"
Client says: "No."
│
┌─────────────────┴─────────────────┐
│ │
Agent typed "No" Agent typed "Yes"
into the system into the system
│ │
▼ ▼
Policy issues at standard rate Underwriting sees it,
refers, loads 50% or
excludes diabetic
complications
│ │
▼ ▼
Client is diagnosed three years Client is covered,
later, files a claim and told plainly why
│ the premium is higher
▼
Carrier investigates medical records
│
▼
Non-disclosure identified
│
▼
Contract avoided. Claim declined.
Policy cancelled. Premium forfeited.
Client has cover for NOTHING.
Agency has a mis-sale file.
The second branch is worse for the client than the first, and that is why the disclosure questionnaire is designed to be answered by the client, in their own words, on their own screen.
This is not distrust of the agent. It is that the whole legal structure rests on the client's own statement. The moment the agent becomes the channel for the answer, the structure has no foundation under it.
Coaching Disclosure Without Coaching the Answer
In practice: the tension in a disclosure conversation is real. Help the client understand what each question means and they will answer it accurately. Steer them toward an answer and you have created the exact problem the disclosure regime exists to prevent.
The legitimate moves are all about recall and comprehension:
| Legitimate | Not legitimate |
|---|---|
| "呢條問題係問你有冇曾經被醫生確診過甲状腺問題，包括甲狀腺功能低下。" | "你應該唔算有甲状腺病喎，係咪？" |
| "如果你只係做過一次體檢，話醫生話唔需要跟進，通常唔算確診。" | "呢個應該唔使填㗎。" |
| "呢度問嘅係住院，門診唔計。回想一下過去五年有冇住過院。" | "你冇住過院啦，係咪？" |
| "唔肯定就唔好亂填，我哋可以安排補充文件核保。" | "填『否』就算啦，差唔多㗎。" |
The right-hand column is where mis-sales are born. Each one is helpful in the moment, and each one is a file waiting to happen.
When a client is genuinely unsure, the correct output is not a guess. It is a referral to supplementary underwriting — get the records, let underwriting decide. That path takes longer and it is the honest one.
type DisclosureAnswer = "YES" | "NO" | "UNSURE";
interface QuestionnaireRule {
questionId: string;
// UNSURE must route to supplementary underwriting, never default to NO.
// This is the single most important invariant in the disclosure module.
unsureAction: "REFER_SUPPLEMENTARY_UW";
// A disclosed condition must never be silently dropped from the summary.
onYes: "ADD_TO_DISCLOSURE_LIST" | "REQUIRE_DECLARATION_ONLY";
}
The Three Underwriting Paths
In practice: not every application goes through the same machinery. The path is chosen by rules — sum assured, product, age, occupation class, and the carrier's own appetite at that moment.
| Path | When it applies | What happens | Turnaround | Client experience |
|---|---|---|---|---|
| Standard | Routine life, medical, savings; sum assured below the auto-approval ceiling; no flags | Rule engine evaluates. Pass or automatic load. No human touches it. | Minutes to 1 day | "批咗" |
| Simplified | Mid-range; declaration-only health Q&A, no medical records required | Rule engine, fewer questions, agent-attested | 1–3 days | "等兩三日" |
| Full | High sum assured, complex occupation, disclosed conditions, or anything the rules can't price | Medical records requested, human underwriter reads the file | 5 days to 4+ weeks | "要等，公司話正常" |
The failure mode an agent must avoid is managing the client to the wrong path. If you tell a Full-underwriting client "usually two days", the day six is a complaint. The path is knowable at submission, so the expectation is settable at submission.
Turnaround expectations, stated at the right moment:
STANDARD → "通常即日或者聽日就有覆"
SIMPLIFIED → "一般兩至三日"
FULL → "要攞醫療報告，等核保同事睇，通常一個星期左右，
複雜嘅個案會耐啲。我一有更新就即刻同你講。"
Never: "好快㗎" — it has no number, so it cannot be measured against.
The last line matters more than it looks. An expectation with a number attached can be met or missed and corrected. An expectation without one can only ever be a complaint.
The Five Outcomes in Commercial Language
In practice: the outcome arrives as an underwriting decision note. The agent's job is to translate it into a sentence with a number in it and a consequence attached.
| Outcome | What the note says | What to tell the client | What it really means commercially |
|---|---|---|---|
| Standard | "Approved at table rate" | "你喺標準條款，冇加費，保障同原本一樣。" | Nothing changed. Say so plainly so they stop worrying. |
| Loaded | "Loading 30% for hypertension, 10-year review" | "加費 30%，保障金額完全冇變，每年保費會貴 X 蚊，呢個係永久寫入張保單。" | The benefit is intact. The problem is affordability over the whole term. |
| Excluded | "Approved with exclusion of IBD-related conditions" | "住院保障會保，但同克隆氏症／腸胃相關嘅住院同手術唔保。呢個係永久嘅，之後加錢都唔可以取消。" | The client may be paying for a benefit that will never be claimed. |
| Referred | "Referred to underwriter — pending medical records" | "已經交咗畀核保同事，佢哋會用你嘅病歷再評估。可能出四個結果：標準、加費、除外或者拒保。我 X 日內同你 update。" | Not a soft approval. Do not imply otherwise. |
| Declined | "Declined — mortality exceeds appetite" | "呢間公司唔接呢個風險，唔係因為你個人有問題，係佢哋喺呢類風險上有自己嘅界線。" | Always look for another carrier. See Lesson 07. |
The three sentences that must be said
Every outcome conversation contains these, in some form:
- What is covered — "你嘅保障金額完全冇變" / "呢一項唔保"
- What it costs or what is excluded — a number, not an adjective
- Is it permanent — "永久寫入張保單" / "之後加費都唔可以取消"
An outcome explained without the third sentence will be re-raised by the client in year four. The permanence is not a warning to soften. It is a term of the contract, and the client signed it.
Evidence, Medical Records, and the Two Rework Loops
In practice: turnaround is decided almost entirely by whether the agent assembles the right documents the first time. Two failure loops account for most of the delay seen by agents, and both are avoidable at the desk.
Loop 1: the missing-referral loop
Agent submits referral → Underwriter: "Please provide specialist report
from the named consultant"
│ │
▼ ▼
Agent asks client Agent did not capture the
for the report consultant's NAME at referral
│ │
▼ ▼
Client does not Agent must go back to the
have it (old records client anyway, to ask
are often destroyed) "who was your doctor?"
│ │
└──────────────┬───────────────┘
▼
Re-referral, +5 to +10 days
Fix: capture the treating clinician's name, the date of last review, and the facility at referral time, not when the underwriter asks. The referral form should refuse to submit without them.
Loop 2: the contradicting-evidence loop
Client discloses: "Asthma, well controlled, no hospital admissions in 5 years"
│
▼
Agent requests medical records ─────┐
│ │
▼ │
Records show: 3 hospital admissions │
for asthma in the last 2 years │
│ │
▼ │
Underwriter: "The disclosure appears │
inaccurate. Please clarify with │
the client." │
│ │
▼ │
┌───────────────────────────────────┴──────────────────────────────┐
│ The trap: the agent "corrects" the disclosure on the client's behalf │
│ to match the records. │
│ │
│ This is now a fresh misrepresentation, and it is worse than the │
│ original non-disclosure, because the agent is now a knowing party. │
└──────────────────────────────────────────────┬──────────────────┘
▼
Correct handling:
✎ Ask the client. Record verbatim.
→ Supplement the original disclosure.
→ Route to supplementary UW.
→ Never overwrite a disclosure entry.
Disclosures are append-only. The correction path is a new entry linked to the original, never an edit. This is not bureaucracy; it is the difference between an accident and a fabrication in the file.
interface DisclosureEntry {
id: string;
questionId: string;
answer: "YES" | "NO" | "UNSURE";
// Free-text verbatim from the client. Not the agent's summary.
clientStatement: string;
// Append-only: a correction is a new entry referencing this one.
supersedes?: string;
recordedAt: string;
recordedBy: "CLIENT" | "AGENT_ASSISTED";
// The agent may facilitate. The agent may not originate an answer.
originatedByClient: boolean; // ← invariant, always true
}
Underwriting in the POS: Where It Lives
In practice: from the agent's point of view, underwriting has exactly three touch points, and the POS must make all three fast.
Touch point 1 — the quote. The quote is a pre-underwriting hypothesis. It is priced on the declared data. The POS should show it as conditional, and never let the agent present it to the client as settled. The minimum a client can be told is: "呢個係按你 而家講嘅資料計嘅,最後批核可能會有分別"。
Touch point 2 — the referral. The referral is a structured data payload, not a free text box. Every field an underwriter will ask for must be captured at the point of referral, because the client may not be reachable later.
{
"referralId": "RW-2026-04412",
"policyRef": "POL-DRAFT-8813",
"outcomeRequested": ["MORTALITY", "MORBIDITY", "CLASS"],
"declaredDisclosures": [
{
"questionId": "COND-THYROID-01",
"answer": "YES",
"clientStatement": "甲狀腺手術 2023 年，之後冇覆診",
"recordedAt": "2026-09-14T10:22:00+08:00",
"recordedBy": "CLIENT"
}
],
"requiredEvidence": [
{ "type": "SURGICAL_REPORT", "clinic": "…", "date": "2023-06", "obtained": true },
{ "type": "LATEST_LABS", "clinic": "…", "date": "2026-08", "obtained": false }
],
"clinicalDetail": {
"treatingConsultant": "陳醫生",
"lastReviewDate": "2023-07-20",
"facility": "…"
},
"expectedTurnaround": { "min": 5, "max": 15, "unit": "BUSINESS_DAYS" },
"clientTold": {
"possibleOutcomes": ["STANDARD", "LOADED", "EXCLUDED", "DECLINED"],
"updateCommitment": "有更新即刻通知"
}
}
Touch point 3 — the outcome. The decision note arrives, the POS records it, and the agent works from the outcome card rather than from an email. The outcome card is required to render the three sentences from the previous section — coverage, cost, permanence — before the agent can move the application forward. If the card cannot be filled in, the conversation is not ready.
Common Misconceptions, Corrected
| Misconception | Correction |
|---|---|
| "轉介 = 批咗" | Referral means a human is reading the file. Four outcomes remain possible. |
| "加費會過幾年消失" | A loading is written into the contract and appears on every premium notice until the policy ends. |
| "除外承保可以之後加錢取消" | It cannot. An exclusion is permanent from inception. |
| "加費代表買咗平啲嘅保單" | The benefit is identical. Only the price changed. |
| "冇住院就唔需要睇醫生" | Outpatient chronic management materially changes morbidity pricing. |
| "客戶唔記得病歷，填否就算" | Unsure routes to supplementary underwriting. Guessing is how contracts get avoided. |
| "披露問錯一題唔會有後果" | Material non-disclosure can void the contract. The claim then fails and the cover did not exist. |
| "如果話加費就一定係有問題" | Loading is a price. Many clients rationally accept it. Some should not — that is an affordability conversation, not a health one. |
A Complete Referral, End to End
In practice: one client, every step, no gaps. This is the shape of a file that comes back in five days instead of four weeks.
CLIENT: 王先生, 41 歲, 非吸煙, BMI 24, 文員, 想買 HK$1,500,000 定期人壽
QUOTE: Standard terms, HK$8,400/yr (hypothesised on declared data)
PATH: Full underwriting (sum assured above auto-approval ceiling)
STEP 1 CAPTURE AT DESK — before anything is submitted
✓ Treating consultant name ✓ Date of last review
✓ Facility ✓ Full disclosure list
✓ Client's own statement for each disclosed item
⚠ Do NOT submit with any of the four missing — the form blocks.
STEP 2 REFERRAL SENT with the structured payload above.
Client is told, in these words, that four outcomes are possible
and that an update is promised by a stated date.
STEP 3 EVIDENCE
D1 Surgical report 2023 — obtained, complete
D2 Latest thyroid panel 2026-08 — obtained, complete
D3 Client did not attend the 2026 follow-up. Honest gap.
→ Agent asks why. Answer recorded verbatim.
→ Routed to supplementary UW. NOT guessed either way.
STEP 4 DECISION NOTE ARRIVES
"Loading 50% for treated thyroid carcinoma, 5-year review,
no exclusion. Declined above HK$2,000,000 sum assured."
STEP 5 AGENT TRANSLATES — three sentences, mandatory
Coverage : "保障金額完全冇變，HK$1.5M 照計。"
Cost : "加費 50%,每年保費 HK$12,600,係永久嘅。"
Permanent: "每張保費單都會見到，之後唔會自己消失。"
STEP 6 CLIENT ASKS THE OBVIOUS QUESTION
"咁加費之後貴咗 $4,200 一年,值唔值?"
STEP 7 AGENT DOES NOT ANSWER WITH A SALE PITCH
→ Refers to the FNA affordability check (Lesson 04).
→ Notes that the sum is capped at $2M by this carrier's appetite.
→ Suggests a second carrier quote (Lesson 07) for comparison.
→ Records the client's decision, whatever it is.
STEP 8 If accepted: new proposal version, superseded original,
fresh signature. Never an edit (Lesson 08).
Step 7 is the one agents skip and the one that creates complaints. The client asked an affordability question. Answering it with a benefits summary is a category error, and the client can feel the difference even if they cannot name it.
When the Outcome Is a Decline
In practice: a decline is the moment where an agent either loses the client or keeps them, and the difference is almost always whether the agent treated the decline as an event about the client or as information about the carrier.
| Do | Don't |
|---|---|
| "呢間公司喺呢類風險上有自己嘅界線，唔係你嘅問題。" | "你身體真係唔夠好。" |
| Offer to quote another carrier. | Treat it as the end of the conversation. |
| Explain what the carrier's appetite boundary usually is. | Speculate about medical reasons you were not told. |
| Record the decline and reason for the comparison record. | Drop the carrier from the file silently. |
| Check whether a reduced sum assured or a different product clears it. | Leave the client with nothing. |
A decline recorded in a comparison file is not a wasted quote. It is one of the most persuasive pages in a bid sheet, because it demonstrates that the search was real. And in this market — multi-carrier by structure, not by choice — it is frequently simply the first of four.
Key Takeaways
- Underwriting decides one thing: whether the carrier takes this risk, on what terms. It resolves on three independent axes — mortality, morbidity, and occupation class — and a carrier can reach different conclusions on each. The agent does not decide. The agent makes the decision legible to a client who just found out the product they bought at 3pm is not the product they get at 6pm. That is a translation skill, and it is the entire module.
- Loaded, excluded, referred and declined are three different things, not degrees of bad luck. Loaded = cover intact, price up. Excluded = the client's most likely claim is carved out. Referred = a human is reading the file and four outcomes remain.
- Both adverse outcomes are permanent, and both must be said out loud. A loading is written into the contract and appears on every premium notice until the policy ends — never tell a client it will go away. An exclusion carves out the condition most likely to generate the claim. A general exclusion (war, riot) applies to everyone; a pre-existing condition exclusion applies to this client only and names the condition. That asymmetry is the sentence that decides whether a client feels informed or feels betrayed in year four.
- Material non-disclosure can void the contract. Not raise the premium — void it. The claim then fails and the cover did not exist. This is why the disclosure questionnaire is answered by the client, on their own screen, in their own words.
- Coach comprehension, never the answer, and never edit a disclosure. Explaining what a thyroid question is asking is legitimate; saying "you probably don't count, right?" creates the exact file the disclosure regime exists to prevent. UNSURE routes to supplementary underwriting, never to a default NO. Disclosures are append-only: the agent may facilitate, but may not originate an answer. An agent-corrected disclosure is a worse finding than the original non-disclosure, because the agent is then a knowing party. A client's later correction is recorded as a new entry, never an overwrite.
- Set the turnaround expectation with a number at submission, using the path the rules actually chose. "Usually two days" for a Full-underwriting case guarantees a complaint on day six.
- Two rework loops dominate delay: the missing-referral loop (capture clinician, last review date and facility at referral time) and the contradicting-evidence loop (record the client's verbatim correction as a new entry, never overwrite a disclosure).
- Every outcome conversation carries three sentences: what is covered, what it costs or what is excluded, and whether it is permanent. An outcome explained without the third will be reopened in year four.
- A decline is information about the carrier, not a verdict on the client. Record it in the comparison file, offer the next quote, and check whether a reduced sum assured clears the appetite boundary.
Glossary (zh-Hant-HK)
| Term | 中文 | Note |
|---|---|---|
| Underwriting | 核保 | The decision to accept the risk and on what terms |
| Underwriter | 核保員 | The human who decides referred cases |
| Pre-authorisation | 預先授權 | Insurer confirms cover before treatment |
| Morbidity | 發病率 | Likelihood and cost of illness |
| Mortality | 死亡率 | Likelihood of death during the term |
| Class / occupation class | 職業風險級別 | Carrier-specific insurability of the job |
| Loading | 加費 | Permanent premium increase written into the contract |
| Exclusion | 除外承保 | Condition carved out of the benefit |
| General exclusion | 一般除外條款 | Applies to every policyholder (war, riot, nuclear) |
| Pre-existing condition exclusion | 既往症狀除外 | Applies to this client only; names a condition |
| Declaration | 聲明／披露 | The client's own statement of material facts |
| Material fact | 重要事實 | A fact a reasonable insurer would price differently |
| Duty of disclosure | 誠實披露責任 | Uberrimae fidei — of the utmost good faith |
| Non-disclosure | 遺漏／不實披露 | Can void the contract, not just raise the premium |
| Rescission | 撤銷合約 | The contract is unwound as if it never existed |
| Simplified underwriting | 簡易核保 | Declaration-only, no medical records |
| Full underwriting | 體檢核保 | Medical records required, human decision |
| Referral | 轉介核保 | File leaves the automated lane for a human decision |
| Standard terms | 標準條款 | Accepted at the table rate |
| Underwriting guideline | 核保指引 | The carrier's own appetite boundary by condition |
| Appetite | 承保意願 | How much of a risk the carrier will take |
| Re-rate | 重新計價 | Recalculate premium after a data correction |
| Evidence pack | 證明文件包 | Surgical reports, labs, specialist letters |
| Specialist report | 專科報告 | Required when a condition is under treatment |
| Turnaround time (TAT) | 核保時間 | Days from submission to decision |
| Supplementary underwriting | 補充核保 | Additional evidence requested after an UNSURE answer |
| Loaded vs excluded | 加費 vs 除外 | Price up vs benefit removed — different problems |
| Appetite boundary | 承保上限 | Above it, the carrier declines regardless of health |