Loan Sizing Engine

Required

Optional (defaults applied if absent)

Process

Step 1: Cash Flow Normalization (T-12 to Lender NCF)

Build the normalization table with three columns: Borrower T-12, Lender Underwritten, Adjustment Notes.

Revenue normalization:

• GPR: compare in-place rents to market. If in-place exceeds market by >5%, underwrite at market for leases rolling within 3 years. Flag above-market leases.
• Vacancy/credit loss: apply the higher of actual vacancy or underwriting floor by property type:
  • Multifamily: 5% minimum
  • Office: 10% minimum (higher for single-tenant or heavy rollover)
  • Retail: 7% minimum (anchored), 10% (unanchored)
  • Industrial: 5% minimum
  • Hotel: use trailing occupancy, stress by 5%
• Other income: normalize non-recurring items (one-time fees, insurance proceeds). Cap percentage-of-revenue items at market norms.

Expense normalization:

• Property taxes: reassess to acquisition basis. Apply local millage rate to purchase price.
• Insurance: trend to current market rates (15-25% annual increases in many markets).
• Management fee: floor at 4% of EGI (multifamily), 3% (office/retail), regardless of self-management claims.
• R&M: normalize to market. Minimum $750/unit (MF), $1.50/SF (commercial).
• Capital reserves (critical -- this converts NOI to NCF):
  • Multifamily: $250/unit minimum
  • Office: $0.25/SF minimum
  • Retail: $0.20/SF minimum
  • Industrial: $0.15/SF minimum
  • Hotel: 4% of revenue (FF&E reserve)

NCF derivation:

NCF = NOI - Replacement Reserves

This is the number the loan is sized against. Not NOI.

Step 2: Loan Sizing Matrix

Size the loan against three simultaneous constraints. Maximum loan = minimum of the three.

Identify the binding constraint (the one producing the lowest max proceeds). This is the constraint that limits the loan.

Step 3: Rate Sensitivity Grid

Key insight: Debt yield is rate-independent. The DY column stays constant across all rate scenarios. As rates rise, the DSCR constraint tightens while DY remains unchanged. At some rate, DSCR becomes binding over DY.

Step 4: Reserve Schedule

Net proceeds = Gross loan - upfront holdbacks. Report both.

Step 5: Rating Agency vs. Originator Gap

Rating agencies typically:

• Apply 5-15% NCF haircut (higher vacancy, lower rents, higher expenses)
• Use stressed cap rates (originator cap + 50-150 bps)
• Use stressed debt constants (higher than actual coupon)
• Result: agency LTV is higher and DSCR is lower than originator underwriting

Quantify the divergence. If agency LTV exceeds 80%, flag potential credit enhancement issues.

Step 6: B-Piece Risk Assessment

Evaluate and assign severity (Low / Medium / High / Deal-Breaker):

• Single-tenant concentration: >50% of revenue from one tenant = High
• Lease rollover concentration: >30% of revenue rolling in years 1-3 = Medium-High
• Tertiary market: outside top-50 MSA = Medium
• Property condition: deferred maintenance, age >30 years without renovation = Medium
• Sponsor track record: limited CRE experience, prior defaults = High
• Franchise/flag risk (hotel): weak flag, franchise expiration during term = High
• Pooling eligibility: loan > 10% of pool = concentration risk flag
• Environmental: Phase I recommendations for further investigation = Medium-High

Step 7: Execution Comparison (if applicable)

Output Format

Present results in this order:

1. Property & Loan Summary -- single-row table with key metrics
2. Cash Flow Normalization Table -- Borrower T-12 vs. Lender UW with adjustment notes
3. Loan Sizing Matrix -- three constraints with binding constraint identified
4. Rate Sensitivity Grid -- base through +200 bps with DY constant annotation
5. Reserve Schedule -- all reserves with upfront holdback and net proceeds
6. Rating Agency Gap -- originator vs. agency with delta
7. B-Piece Risk Flags -- severity-rated checklist
8. Execution Comparison -- multi-lender comparison (if applicable)

Red Flags & Failure Modes

1. Sizing off NOI instead of NCF: The single most common error. Replacement reserves must be deducted before sizing. A $250/unit reserve on 200 units = $50K/year difference in max proceeds at 1.25x DSCR = $600K+ in loan sizing.
2. Using borrower vacancy instead of underwriting floor: In-place 97% occupancy does not mean 3% vacancy for sizing. Apply the property-type floor.
3. Tax reassessment omission: Property taxes reassess to acquisition basis in most jurisdictions. Failing to adjust understates expenses and overstates NCF.
4. Ignoring reserve holdbacks: Gross proceeds and net proceeds can differ by $500K-$1M+ after upfront holdbacks. Always report both.
5. Confusing DY with DSCR: Debt yield is rate-independent. It measures income relative to loan balance regardless of coupon. DSCR measures income relative to debt service, which depends on rate. These constraints bind at different rate levels.
6. Above-market lease reliance: A lease at 130% of market rent expiring in year 2 creates a roll-down risk that agency underwriting will capture even if originator underwriting does not.

Chain Notes

• Upstream: rent-roll-analyzer (pre-normalized rent roll), deal-underwriting-assistant (equity-side needs debt inputs)
• Downstream: mezz-pref-structurer (senior sizing determines gap), capital-stack-optimizer (senior is first input), refi-decision-analyzer (sizing methodology for maturity analysis)
• Peer: sensitivity-stress-test (rate sensitivity methodology shared)

Computational Tools

This skill can use the following scripts for precise calculations:

• scripts/calculators/debt_sizing.py -- sizes loan against simultaneous DSCR, LTV, and debt yield constraints with rate sensitivity grid

  python3 scripts/calculators/debt_sizing.py --json '{"noi": 1500000, "property_value": 20000000, "target_dscr": 1.25, "target_ltv": 0.65, "target_debt_yield": 0.09, "rate": 0.065, "amortization_years": 30, "io_years": 2}'