Bridge Loans to Permanent Debt: Timing the Refi in REPE Models

Stabilized multifamily often closes on agency or life company debt from day one. Value-add and lease-up deals usually cannot. In-place NOI is too low for the LTV and DSCR the permanent lender wants, so sponsors use a bridge loan, execute the business plan, then refinance into cheaper permanent debt. Interviews love this sequence because it ties debt sizing, occupancy timing, and exit (or hold) returns together.

Why bridge exists

Bridge lenders underwrite the business plan, not just today's cash flow. They care about:

• In-place coverage (often interest-only for a period)
• Future stabilized NOI
• Sponsor equity and experience
• Capex budget and leasing velocity
• Exit or refi as the takeout

Permanent lenders (agency, CMBS, life cos) want a stabilized asset: occupancy covenant, DSCR test on trailing or underwritten NOI, tighter LTV. If you are at 78% occupancy with a renovation underway, you are not getting Fannie Mae pricing yet.

Typical capital path

Close: Bridge loan + equity fund purchase + capex
Operate: Lease-up, renovate, mark rents to market
Stabilize: Hit occupancy / NOI / DSCR hurdles
Refi: Pay off bridge with permanent loan (and maybe pull equity)
Hold or sell: Permanent debt stays until disposition

Some deals skip the hold and sell at stabilization. The refi math still matters because a buyer will size their own permanent debt off forward NOI.

Bridge terms you should know cold

Floating rate plus short term is the point. You are paying for flexibility while NOI ramps. You are not trying to lock a 10-year fixed coupon on a half-empty building.

The refi sizing problem

At refi, permanent loan amount is usually the minimum of:

Max Loan (LTV)  = Max LTV × Stabilized Value
Max Loan (DSCR) = Stabilized CFBD ÷ (Min DSCR × Debt Constant)

Stabilized value itself is often:

Stabilized Value = Stabilized NOI ÷ Cap Rate (or appraisal)

If bridge principal is $70M and the permanent lender only sizes to $62M, you have an $8M hole. Equity fills it, or you sell, or you extend the bridge (expensive). This is the trap interviewers plant: your operating model hits 95% occupancy, but you never checked whether the takeout loan covers the bridge balance.

Which constraint binds

• LTV-constrained takeout: value is the limiter; NOI may look fine
• DSCR-constrained takeout: cash flow is the limiter; appraisal value may look fine

Say which one binds. It changes the story about risk.

Timing mismatches that break models

Stabilization date vs loan maturity. If lease-up finishes in month 30 and the bridge matures in month 24, you need an extension assumption or a faster plan.

NOI used for DSCR. Permanent lenders may want trailing 3 months annualized or a full T-12 after stabilization, not your "month 1 of stabilized" forecast. Models that refi the day occupancy hits 95% are optimistic.

Capex still outstanding. Lenders haircut value or require reserves if renovations are incomplete.

Rate shock. Floating bridge rates that jump 200 bps can wipe Year 1 cash flow even if the lease-up is on track.

Loss-to-lease still open. Stabilized occupancy with in-place rents still below market can disappoint permanent sizing if the lender uses in-place NOI.

Simple interview example

Purchase $100M, 65% bridge ($65M) at SOFR+3.5%, interest-only. Stabilize Year 2 at $7.0M NOI. Cap for value = 5.5% → value ≈ $127M. Agency will do 70% LTV and 1.25x DSCR at a 6.5% debt constant.

LTV max   = 0.70 × $127M = $89M
DSCR max  = $7.0M ÷ (1.25 × 0.065) ≈ $86M
Permanent = min ≈ $86M

You can pay off the $65M bridge and pull ~$21M of equity at refi (before fees and reserves). That equity pull is a cash flow in Year 2. It boosts IRR even if MoIC at final exit barely changes. Same MoIC vs IRR lesson as in PE: timing matters.

Extension downside variant

If stabilization slips and you extend one year at a 1% fee plus higher rate, common equity IRR can fall hard even if the eventual NOI target is unchanged. Interviewers like that stress.

When you would not use bridge

• Asset is already stabilized and qualifies for agency/life company terms
• Cap rates and NOI support the leverage you need at close
• Sponsor does not want floating rate risk
• Business plan is light and a smaller senior loan plus more equity is cleaner

30-second answer

"Bridge is transitional capital sized on the plan. Permanent is takeout capital sized on stabilized NOI and value. I model the bridge through lease-up, then size the permanent loan on LTV and DSCR at stabilization, and I check that the takeout covers the bridge balance. If it does not, I need more equity, a lower purchase price, or a longer bridge with extension cost in the case."

Interview prompts

"When would you use a bridge loan?"
"When in-place NOI cannot support permanent LTV/DSCR at close, but the business plan should get there. Bridge funds the transitional period; permanent is the takeout."

"What is the biggest risk in a bridge-to-perm structure?"
"Stabilization timing and takeout size. If permanent debt cannot fully refinance the bridge, you need more equity, an extension, or a sale into a weaker market."

"How do floating rates change the case?"
"Bridge interest can spike and crush Year 1 cash flow even if leasing is on track. I stress SOFR up and check IO coverage."

"Does a successful refi always mean a great deal?"
"No. A refi that barely covers the bridge with no equity pull can still leave a mediocre MoIC if you overpaid at entry. Refi solvency and sponsor returns are related but not identical."

Practice checklist

1. Size bridge at close on cost/value and coverage
2. Build lease-up to stabilized NOI
3. Size permanent on LTV and DSCR; take the min
4. Confirm takeout ≥ bridge (or quantify the hole)
5. Model an equity pull if takeout exceeds bridge
6. Shock exit timing +12 months with extension fee
7. Shock floating rate +200 bps on Year 1

Run a value-add acquisition where Year 1 is bridge IO and Year 2 (or 3) is a refi into amortizing permanent debt. Force the takeout to be DSCR- or LTV-constrained so you see which one binds. Then shock timing by 12 months and recompute extension cost. That is the skill that shows up in real IC memos.