The Loan After the Loan
Every construction loan has an expiration date. Developers spend months negotiating rate, leverage, and covenants on the construction facility, then treat the exit as an afterthought. That is backwards. The take-out loan, the permanent financing that retires construction debt at stabilization, is where a deal’s actual return gets decided. Hotel construction refinance risk is not a closing-day problem. It is underwritten, or ignored, at the term sheet stage, long before the first shovel hits the ground.
If you are modeling IRR off a construction-to-perm assumption that no longer reflects the market, the pro forma you raised capital on is already stale.
How Take-Out Financing Works
Construction loans are short-duration and priced for a project that does not yet generate cash flow. Once the hotel opens and stabilizes, typically after ramping trailing twelve-month NOI to a level a permanent lender will underwrite, the borrower refinances into a longer-term loan sized against actual performance rather than pro forma projections. That permanent loan pays off the construction facility.
The gap between those two events is where risk concentrates. Permanent lenders underwrite to a debt yield or debt service coverage ratio based on trailing NOI, not the NOI in your original feasibility model. If the asset stabilizes below projection, or takes longer to stabilize than modeled, the loan amount the take-out lender is willing to fund shrinks. That gap has to be filled with additional equity, mezzanine debt, or an extension of the construction loan at a higher rate and stricter terms.
Schedule performance during construction directly shapes this outcome. A project that opens on budget and on time preserves the ramp period assumed in the pro forma. A project delayed by schedule drift compresses the stabilization window against the same maturity date, arriving at refinance with a lower trailing NOI than modeled. The mechanics of that erosion are covered in detail in our piece on how schedule drift erodes projected IRR.
Rate Environment Risk
Construction loans typically run two to three years. Nobody can forecast where permanent debt will price at maturity. That uncertainty is not a rounding error in the model, it is a variable that determines whether the deal clears its return hurdle at all.
Consider the mechanics. If the construction loan was underwritten at a permanent takeout rate of 6.5 percent and the market clears at 8 percent instead, the same NOI supports a materially smaller loan balance under a fixed debt yield constraint. That shortfall lands on the sponsor’s balance sheet as a cash-in refinance, a bridge extension, or a forced capital call to LPs who did not sign up for it. Refinance risk compounds when rate risk and construction delay risk hit at the same time, because a delayed opening pushes your refinance date into whatever rate environment exists later, not the one you modeled at closing.
This is why sensitivity analysis matters more than a single base-case IRR. A feasibility model that only shows returns under one refinance rate assumption is not a risk assessment, it is a hope. Sponsors who stress-test the take-out across a range of rate and NOI scenarios go into capital raises with a defensible answer when an LP asks what happens if rates are still elevated at stabilization.
Structuring for Flexibility
The tools for managing this risk exist upstream of the refinance date, not at it. A few structural choices matter most:
- Extension options built into the construction loan, negotiated at origination, buy time if stabilization or the rate market runs long. Negotiating these after the fact, with a lender who knows you have no leverage, costs far more.
- Rate caps or hedges on floating-rate construction debt limit downside exposure during the build period, when the project has no revenue to absorb rate shocks.
- Schedule certainty protects the stabilization window the take-out lender is underwriting against. This is where general contractor selection intersects with capital markets risk. A GC involved early enough to catch design and constructability issues before they become field change orders protects the delivery date the refinance depends on, a connection explored in how early GC involvement protects the capital stack.
- Conservative NOI assumptions in the original pro forma reduce the gap between projected and trailing performance at refinance, narrowing the range of outcomes a lender has to underwrite around.
None of this eliminates hotel construction refinance risk. It converts an unmanaged variable into a modeled one, which is the difference between a capital call that surprises your LPs and one they anticipated in the offering documents.
Value engineering decisions made during design also ripple forward into this calculus. A construction budget that was never stress-tested against the design intent tends to surface cost overruns mid-build, extending the schedule and pushing the refinance date later than planned. That chain of causation starts long before groundbreaking, as outlined in why budget overruns start in design. Similarly, a construction bid that looks attractive because it is the lowest number on the table often carries embedded risk that surfaces as change orders and delay, the exact dynamic examined in why the lowest bid is a risk signal, not a value.
Underwrite the Exit Before You Underwrite the Entry
Take-out financing is not a formality to handle when the hotel opens. It is a second underwriting event, subject to its own rate environment and its own NOI test, and it deserves the same scrutiny at the term sheet stage as the construction loan itself. Sponsors who model refinance risk into their capital stack from day one negotiate better extension terms, size their equity more accurately, and avoid the kind of capital call that damages an LP relationship for the next three deals.
If your current model treats the take-out as a single-point assumption rather than a range of outcomes, a pro forma and feasibility review is the place to start closing that gap before construction debt is drawn.