The Budget That Doesn't Fund: How Lenders Underwrite GMP vs. Cost-Plus, Contingency, and Change-Order Governance in Ground-Up Construction Loans

Every ground-up construction loan begins with a fantasy: a budget that will be executed as written. Lenders know better. What separates a clean close from a three-year nightmare is not the accuracy of the budget, but the governance framework that determines who pays when reality diverges from the plan.

The structural constraint here is ancient and immutable. Construction is the only major asset class where the final product's cost is unknown at origination. Appraisers value finished buildings. Lenders fund construction. Between those two moments lies contingency, change orders, and the eternal contest over which party absorbs cost growth.

This dynamic has not changed in fifty years, and it will not change. The budget that funds a loan and the budget that actually gets built are different documents. How lenders underwrite the gap between them is where real differentiation lives.

The GMP Illusion and Its Structural Limits

A Guaranteed Maximum Price contract sounds like what it says. The general contractor guarantees the price. In practice, a GMP is a price subject to documented changes, unforeseen conditions, and force majeure carve-outs that can be fifty pages long.

Lenders know this. What they underwrite is not whether the GMP will hold, but whether the sponsor has structured the GMP to absorb the categories of cost growth that actually occur in ground-up work. A tight GMP with a narrow scope of included work and broad change-order definitions is not a GMP. It is a cost-plus contract wearing a GMP label.

The structural problem is this: a truly tight GMP (one that covers unforeseen conditions, soil remediation, code changes, and design modification) makes the contractor uncompetitive in bidding. Contractors bid loose GMPs and negotiate change orders later. Sponsors and lenders know this happens. They price for it anyway, because the alternative, full cost-plus, signals to the market that cost control is absent.

Cost-Plus Transparency and the Hidden Tax

Cost-plus contracts reverse the problem. Every cost is documented, submitted, and approved. Transparency is real. But cost-plus introduces its own structural constraint: the contractor has no incentive to optimize labor efficiency, waste reduction, or schedule compression. Cost growth becomes a feature, not a bug.

Lenders underwriting cost-plus deals build in a cost inflation assumption that exceeds the GMP contract's stated contingency. This is not conservatism. This is historical pattern recognition. Over decades, cost-plus projects run 8-15% above the initial budget line item. GMP projects run 5-12% above. The difference is not the contract type. It is that cost-plus sponsors have already accepted cost growth as inevitable, so they staff, bid, and negotiate with less discipline.

The structural insight: cost-plus is not cheaper than GMP. It is more honest about where cost lives. Lenders price accordingly, and smart sponsors know that a cost-plus underwrite will produce a tighter contingency reserve than a GMP underwrite, because the lender is already assuming growth in the base number.

Contingency as a Governance Tool, Not a Safety Net

A construction budget typically allocates 5-15% of hard costs to contingency. Lenders do not view this contingency as available capital. They view it as the boundary between sponsor-absorbed risk and lender-absorbed risk.

Here is the structural constraint: contingency that is not spent does not revert to the lender. It reverts to the sponsor. Therefore, sponsors have an incentive to spend contingency before project completion, and contractors have an incentive to allocate costs to contingency rather than base budget (where they might trigger cost overruns that breach the GMP). Lenders know this. They underwrite contingency as a governance tool: they require change-order documentation for any draw against contingency above a certain threshold, they tie contingency release to specific project milestones, and they reserve the right to refuse contingency draws that appear to subsidize poor scheduling or cost management.

The lender's real underwrite is not the contingency percentage. It is the mechanism that controls contingency release. A sponsor with weak controls and a loose contingency definition will burn through reserves by month twelve. A sponsor with tight controls and clear thresholds can stretch contingency to project completion and absorb real unforeseen costs.

Change-Order Governance and the Release Holdback Cascade

Change orders are the mechanism by which construction budgets expand. A standard loan document requires lender approval of any change order above a threshold (typically 0.5-2% of total project cost). Below the threshold, the contractor and sponsor can approve changes without lender consent.

This creates a structural incentive: break large changes into smaller ones, each just below the approval threshold. A $500K change order that requires lender approval gets bundled into five $90K orders that do not. Lenders know this happens. They mitigate it by requiring monthly change-order schedules, cumulative tracking across the project, and the right to audit the change-order log at any time.

But the real governance mechanism is the release holdback. Lenders typically hold back 10-20% of each construction draw. That holdback is released when the project is completed and the lender has confirmed that no outstanding change orders exist, that the budget was not exceeded, and that the as-built cost matches the as-completed appraisal. If change orders have accumulated beyond the contingency reserve, the sponsor must cover the difference from the held-back funds. If held-back funds are insufficient, the sponsor must inject additional capital or the project cannot be released.

This is the structural mechanism that forces discipline. The sponsor knows that every change order reduces the capital available at the end. Contractors know that lenders will not fund work that was not in the approved budget. The result is that change orders are negotiated ruthlessly, because they have real cost to both parties.

The Underwrite That Separates Winners From Everyone Else

Sophisticated lenders do not underwrite the budget. They underwrite the sponsor's change-order discipline and the contractor's incentive alignment. They ask questions that look simple but reveal everything:

How many change orders does this contractor typically generate on similar-sized projects? What percentage of the budget do those changes represent? How does this sponsor's project manager track contingency usage? Who approves contingency draws, and what documentation is required? What is the escalation path if change orders exceed a cumulative threshold?

These questions sound administrative. They are structural. They determine whether the budget that was underwritten is the budget that will be executed, or whether the sponsor will be funding cost growth from pocket capital by month eighteen.

The budget that funds a construction loan is not the budget in the pro forma. It is the budget that survives the change-order governance framework. Lenders who understand this difference do not have construction surprises. Those who do not, fund them repeatedly.

What I'm watching next

• Whether sponsors are tightening GMP scope to reduce change-order friction, or whether contractors are refusing tighter GMPs and pushing cost-plus structures that shift risk back to lenders via contingency underwriting.
• How construction cost inflation (labor, materials, supply chain delays) is forcing lenders to haircut sponsor contingency assumptions, and whether that is creating a feedback loop where contingency reserve adequacy becomes the binding constraint on loan sizing.
• Whether construction loan documents are evolving to require real-time change-order tracking (blockchain, smart contracts, automated approval workflows) or whether sponsors and contractors are resisting transparency because it makes cost growth visible and reduces negotiating flexibility.