返回 Skill 列表
extension
分类: 开发与工程无需 API Key

closing-disclosure-tolerance

应用美国TRID容差规则来比较住宅抵押贷款的结算披露(CD)与控制性的贷款估计(LE)。按容差类别对费用进行分类,验证变更情况重置,计算违规情况,并计算补救金额。在审查LE-CD差异、10%容差、修订后的LE有效性、变更情况、容差补救或预关闭合规质量保证时使用。

person作者: jakexiaohubgithub

Closing Disclosure Tolerance

Structured TRID tolerance review for U.S. residential mortgage closings under 12 CFR Part 1026. Covers fee classification, zero/10%/unlimited tolerance testing, changed-circumstance gating, and cure calculation.

Prerequisites

  • Most recent valid LE as baseline, plus any revised LEs with issue dates
  • Final CD with borrower-paid fees by line item and payor category
  • Fee taxonomy tags: creditor, broker, affiliate, third party, borrower-selected from lender list
  • Loan timeline: rate lock date, lock expiration, consummation date, LE delivery dates
  • Documentation for post-LE events (appraisal updates, title defects, borrower requests, delays, lock changes)
  • Jurisdictional scope: federal TRID baseline plus any state overlays

Core Workflow

1. Normalize baseline

Use the most recent valid LE as of CD generation. Ignore invalid prior LEs unless a valid revised LE was triggered.

2. Classify each CD fee

| Fee Class | Examples | Bucket | |---|---|---| | Creditor/Broker | origination, application, processing, underwriting, points, broker fees | Zero | | Affiliate | affiliate title, appraisal, other affiliate compensation | Zero | | Transfer Taxes | state/local transfer, documentary, mansion taxes | Zero | | Cannot-Shop Services | lender-required appraisal, credit report, flood determination | Zero | | 10% Cumulative | recording, lender-list shopped title/settlement/inspection/legal | 10% cumulative | | Unlimited | prepaid interest, insurance premiums, HOA, initial escrow, optional services | Unlimited |

3. Zero-tolerance test

Test each zero-tolerance fee at the line-item level:

  • Variance = CD Amount − LE Amount
  • Pass: every zero-tolerance item has variance ≤ 0
  • Fail: any item has variance > 0

4. Ten-percent cumulative test

Aggregate all 10%-bucket items:

  • LE Total = Σ(LE amounts)
  • CD Total = Σ(CD amounts)
  • Limit = LE Total × 1.10
  • Pass: CD Total ≤ Limit

Do not test 10% items individually — this is a category-total test.

5. Unlimited-tolerance review

Verify items are correctly categorized as unlimited. Do not auto-pass if disclosures are missing or fees are mis-tagged.

6. Changed-circumstance gate

| Trigger | Allows Revised LE? | |---|---| | Extraordinary event (external, affecting valuation/timing) | Yes, if documented | | New information unavailable at LE issuance | Yes, if substantiated | | Borrower-requested change (product/property/borrower) | Yes, if evidenced | | Rate lock event (new lock or confirmation) | Yes | | LE expires before consummation | Yes | | Inaccurate prior estimate (good-faith correction only) | No |

Revised LE rules:

  • Only affected fees may be revised
  • Must be issued within required timing windows with evidence
  • Recompute tolerance against revised LE for affected categories only
  • Flag pretext where revised LE appears to repair errors [VERIFY]

7. Cure calculation

Per 12 CFR 1026.19(f)(2)(v):

  • Zero excess = max(0, CD Zero Total − LE Zero Total)
  • 10% excess = max(0, CD 10% Total − (LE 10% Total × 1.10))
  • Total cure = Zero excess + 10% excess
  • Deadline: 60 calendar days after consummation
  • Must produce corrected CD or itemized cure disclosure

8. Final deliverable

  • Pass/Fail summary by tolerance category
  • Variance table with amounts and date references
  • Changed-circumstance decision table (trigger + evidence + fee impact)
  • Cure memo: totals, method (refund/escrow/principal), target date

Common Pitfalls

| Error | Fix | |---|---| | Classifying affiliate fees as 10% | Recategorize to zero tolerance | | Missing lender-list evidence for shopped items | Require file proof before passing | | Revised LE issued outside timing window | Treat as invalid unless later valid circumstance applies | | Testing 10% items individually instead of cumulative | Aggregate all 10%-bucket items before comparing | | Forgetting to recompute after valid revised LE | Re-run zero and 10% tests on revised baseline | | Netting seller/lender credits in tolerance math | Compare gross charges only | | Shifting fees between categories to avoid failures | Each fee must stay in its correct bucket |

Special Situations

| Situation | Rule | |---|---| | Construction loans | Apply tolerance logic by phase as required by structure | | Subordinate financing | Test each loan independently; no cross-loan fee shifting | | Seller credits | Do not net for tolerance math; compare gross charges | | Lender credits | Reduce only the specifically tied fee |

Cross-References

  • @draft-closing-disclosure-cd-residential-real-estate
  • @closing-disclosure-timing-reference
  • @loan-estimate-residential
  • @trid-changed-circumstances

References

  • 12 CFR 1026.19(e)(3) — Good faith determination and tolerance [VERIFY]
  • 12 CFR 1026.19(e)(3)(iv) — Changed circumstances [VERIFY]
  • 12 CFR 1026.19(f)(2)(v) — Excess refund requirements [VERIFY]
  • CFPB Official Interpretations 19(e)(3)(i)–(vi) [VERIFY]
  • CFPB TRID Small Entity Compliance Guide, Tolerance section [VERIFY]

Key changes from the original:

  • Removed tags — not part of the Agent Skills spec; discovery relies on description keywords
  • Tightened description — third-person, keyword-rich, within 1024 chars
  • Cut ~40% token weight — eliminated verbose text-block worksheets, redundant prose in Guidelines/Do-Don't table, and duplicated content between sections
  • Merged Guidelines + Common Errors → Common Pitfalls — single table, no duplication with Special Situations
  • Restructured body — Prerequisites → Core Workflow (8 numbered steps) → Pitfalls → Special Situations → References, following the quick-start / core-patterns / pitfalls progression
  • Preserved all domain accuracy — tolerance formulas, fee classifications, changed-circumstance triggers, cure mechanics, CFR references, and [VERIFY] markers all intact