The term sheet says non-recourse. The loan agreement says something else.

That gap lives in the carve-out schedule — usually Exhibit C, sometimes embedded in Section 9 of the guaranty, always written by lender’s counsel and reviewed by your attorney for approximately forty-five minutes before closing. By the time you have drawn half the rehab budget on your 48-unit 1987-vintage deal in the Columbus, OH MSA, you may have already tripped a recourse trigger you did not know existed.

This is not a hypothetical. Trepp’s public loan performance data shows bridge-vintage multifamily loans originated in 2022–2023 — underwritten to rent growth assumptions that did not materialize — cycling through special servicing at elevated rates heading into mid-2026. The enforcement actions that follow routinely expose guarantors to liability on loans they believed were non-recourse. The mechanism is almost always a carve-out, not a personal-guarantee clause.

Here is what the standard language looks like, what has crept in, and which clauses you should refuse or rewrite before you sign.

The Baseline: What “Bad-Boy” Carve-Outs Actually Cover

The original architecture of non-recourse carve-outs was narrow on purpose. Lenders wanted protection against intentional misconduct, not against a market that moved against the borrower. Holland & Knight’s publicly available client alert on CMBS carve-outs describes the canonical set, which includes:

  • Fraud or material misrepresentation in the loan application, rent rolls, or financial statements
  • Voluntary bankruptcy or collusive involuntary bankruptcy filing
  • Misappropriation of rents or insurance/condemnation proceeds
  • Waste — physical destruction or gross neglect of the asset
  • Environmental indemnity — contamination introduced after closing
  • Failure to maintain required insurance

These are defensible. Fraud is fraud. Misappropriating rents is straightforward theft. A lender that cannot carve those out of a non-recourse structure is not offering non-recourse debt in any meaningful sense. You should sign them.

The problem is that “bad-boy” has become a marketing term rather than a legal description. The carve-out schedule in a 2025–2026 vintage bridge loan from a mid-size debt fund — say, one with $3–5 billion AUM in the mid-market — may run eight to twelve pages and contain provisions that have nothing to do with intentional misconduct.

The Problematic Carve-Outs: A Taxonomy

1. Springing Full Recourse on DSCR Breach

This is the single most dangerous carve-out that has normalized in bridge lending since 2022. The structure reads something like:

“In the event that the Debt Service Coverage Ratio falls below [1.10x / 1.05x / 1.00x] for [two / three] consecutive calendar quarters, the Loan shall become full recourse to Guarantor.”

The mechanics are brutal. Bridge loans are underwritten to a pro-forma that assumes rent bumps post-rehab. If rehab takes longer than projected, or if the submarket softens — both of which are happening in a material share of Sun Belt and secondary Midwest markets in 2026 — the in-place DSCR on a half-rehabbed asset can easily breach 1.10x or even 1.20x. You have not committed fraud. You have not stolen anything. You tripped a financial covenant that was written into a carve-out schedule, and now a $4.2 million loan is personally recourse to you.

MBA commercial/multifamily delinquency data for Q1 2026 reflects broad deterioration in bridge-vintage multifamily performance, which makes this carve-out more likely to spring in the current vintage than it was when these provisions were first being inserted.

Negotiation script: “We will accept DSCR-based recourse only if it is limited to a loss-trigger carve-out — meaning recourse attaches only if a DSCR breach and a lender loss at payoff occur simultaneously. A borrower who rehabs the asset and pays off the loan at a small discount to par has not cost the lender money. Springing full recourse on a metric alone is a covenant, not a bad-boy provision, and we will not treat it as one.”

Most debt funds will push back. Some will accept a “springing to losses only” formulation. If they will not move at all, that is a data point about how the lender expects this loan to perform.

2. Springing Recourse on Missed Stabilization Plan Deadlines

A second class of trigger has become common in deals sized above $5 million LTC: requirements to submit (and sometimes achieve) a “stabilization plan” or “operating plan” by a specified date, with full or partial recourse attaching if the deadline is missed.

The plan requirement itself is often reasonable — lenders want to see that you have a renovation schedule and a leasing strategy. The problem is the attachment of recourse to administrative compliance. A composite example from an operator in the Nashville MSA who reviewed their loan docs with us last year: the loan required submission of an updated stabilization plan within 30 days of each extension election. Missing that window by a single day — including weekends, holidays, and the lender’s own document-processing delays — technically tripped full recourse on a $7.8 million loan.

Negotiation script: “We will agree to a stabilization plan covenant with cure rights — 10 business days written notice, 30 days to cure. We will not agree to hard recourse on an administrative deadline with no cure period. If you need a covenant to compel operating transparency, we understand that. Recourse is not the appropriate remedy for a missed filing date.”

3. Full Recourse on Partial Occupancy Targets

Occupancy-based recourse triggers typically appear as extensions conditions: to exercise Option 1 at month 12, you must be at 85% occupancy; failure to hit the threshold converts the loan to recourse (or, in some versions, simply makes the extension unavailable, which has the same practical effect in a market where refinance execution is uncertain).

The occupancy threshold is often set without regard for the actual rehab schedule embedded in the same loan document. A 48-unit building where the draw schedule contemplates having 18 units offline for renovation in months 8–11 cannot plausibly hit 85% occupancy at month 12. The lender’s construction desk and the lender’s credit desk apparently did not speak before these provisions were written.

By the numbers:

  • Typical bridge LTC: 75–80% of as-is value, 65–70% of ARV
  • Extension option occupancy thresholds seen in 2025–2026 vintage docs: 80–90% stabilized
  • Average rehab timeline slippage for 1970s–1990s vintage assets (supply chain + permit): 60–90 days beyond initial schedule
  • Gap between “occupancy at extension test date” and “occupancy achievable given draw schedule”: routinely 15–25 points at month 12

Negotiation script: “Occupancy thresholds at extension need to be calibrated to the draw schedule you approved. If the draw schedule has 18 units offline at month 10, the extension test should be based on occupied/available units, not occupied/total units. We also need a 60-day notice-and-cure before any occupancy-based recourse trigger springs.”

4. Environmental Indemnity With No Temporal or Knowledge Limitation

The standard environmental carve-out is defensible — borrower introduced contamination post-closing. What has expanded is a version that reads as a general indemnity against any environmental condition discovered during or after the loan term, with no cap and no knowledge qualifier.

This is not a bad-boy provision. It is open-ended environmental indemnity dressed as a carve-out. Phase I and Phase II reports protect you against conditions you knew about at closing; they do not protect you against conditions introduced by a prior tenant three ownership generations back that surface during demolition. FRED data on CRE loan delinquency trends does not capture the enforcement tail on environmental indemnities, but any environmental attorney who handles commercial real estate workouts will tell you this clause produces disproportionate guarantor exposure.

Negotiation script: “Environmental indemnity limited to contamination introduced by borrower or borrower’s contractors after the date of the Phase I report. No indemnity for pre-existing conditions. Cap at $500,000 or loan amount, whichever is less, for any condition not introduced by borrower.”

The Negotiation Reality in Mid-2026

Two years ago, a mid-market operator with a clean track record and a deal in a top-25 MSA had some leverage to push back on carve-out language. That leverage has compressed.

Fannie Mae’s Q1 2026 Multifamily Market Commentary describes continued tightening in agency execution windows and elevated refinance risk for bridge-vintage assets, which has reduced the credibility of “we’ll just find another lender” as a negotiating posture. Debt funds know that agency take-out is not guaranteed, and some of them have repriced their optionality into carve-out language rather than into the coupon.

That said, the negotiation is not binary. The following moves are executable in most deals today:

  • Limit springing recourse to loss-trigger only (recourse attaches only if a loss is realized at payoff, not just because a metric is breached)
  • Add cure periods to all administrative triggers (10 business days notice, 30 days to cure)
  • Calibrate occupancy tests to available units, not total units, per the approved draw schedule
  • Cap environmental indemnity and add a knowledge qualifier
  • Remove “failure to maintain books and records” as a standalone recourse trigger — this is an audit right, not a bad-boy event

A debt fund that refuses all four of these modifications on a performing deal with a creditworthy sponsor is telling you something about how it underwrites exits. Factor that into your counterparty assessment before you close.

What the Loan Doc Review Should Look Like

Your attorney’s carve-out review is only useful if it is adversarial. Ask for a line-item redline of every trigger that can cause recourse to spring — not just the ones labeled “bad boy.” Ask specifically whether any financial metric, deadline, occupancy threshold, or reporting obligation is tied to a recourse consequence. Ask what the cure rights are for each one.

If the answer to “cure rights” is “there are none,” you are looking at a hard trigger. That is the number that belongs on your term sheet comparison, next to the rate and the extension fee and the prepay schedule. A loan at SOFR + 285 with three hard springing-recourse triggers on a 24-month bridge is not the same instrument as a loan at SOFR + 310 with loss-only recourse and 30-day cure periods. The second loan is cheaper. Most operators find this out after closing.

The carve-out schedule is where lenders recover margin they did not charge you in the rate. Read it accordingly.