A model reads a contract for its numbers. The clause that decides the deal often doesn’t have one.
The provision that decided whether the incentive fee was worth millions or worth nothing carried no number. No percentage, no threshold, no dollar figure — just a contingency, buried in the financing terms, easy to miss on a first read. A model built to pull parameters from a fee schedule reads right past it.
Financial models read contracts for numbers. Fee percentages, threshold margins, RevPAR Index minimums, cure caps. The work is mechanical: find the number, put it in the spreadsheet, run the Monte Carlo, print the NPV. For most provisions, that works. Fee structures are numbers. Performance tests are numbers. Cure mechanics are numbers. The model reads them and produces the analysis.
But the clauses that most often flip a deal’s economics don’t carry numbers. They’re carve-outs, contingencies, blank lines left for later negotiation. Call it a no-number clause: a provision that decides which numbered structure applies without holding a number itself. This is incomplete-contract theory in plain form. No agreement can spell out every future contingency, so the gaps get resolved later — and the party positioned to resolve them holds the real power (Hart, 2016). None of this is a defect. Contracts leave gaps because they have to. But a model that only reads numbers walks past the gaps. The NPV prints. The bid lands. And the structure the contract actually creates stays invisible until it surfaces after signing — when the economics can’t be modeled and the negotiation can’t be reopened.
Rigorous HMA analysis needs a contract-reading lens, not just a modeling lens.
What the Model Reads
Standard analysis pulls a defined set of inputs. Fee percentages — base and incentive. Performance thresholds — GOP margin floors, RevPAR Index minimums. Cure mechanics — payment caps, window lengths, the number of cures allowed. Key money and amortization. Term and option periods.
These feed straight into the NPV. The model runs the scenarios, stress-tests the thresholds, weights the outcomes. For most provisions, that’s enough. They’re numbers. The numbers drive the result. The model catches the economics.
It breaks when the provision that decides the structure carries no number. The contingency that forks the incentive into two different configurations. The blank that defers a threshold to a later negotiation. The carve-out whose knock-on effect on consecutive-failure mechanics can’t be read off the clause. These don’t feed a spreadsheet. They decide which spreadsheet to build.
The sharpest case is a tax-status contingency in a tax-exempt bond deal.
The Tax-Status Contingency
Start with a hotel financed with tax-exempt bonds. The incentive has two gates: a GOP margin threshold and a RevPAR Index threshold. A standard read sees a double-gate incentive — both gates have to clear. Run the numbers, and the margin gate clears in most scenarios while the Index gate fails in most. Nothing in reach — ADR lifts, occupancy gains, cost cuts — clears both at once. The incentive appears unearnable. NPV contribution: zero.
But the fee schedule isn’t the whole structure. Elsewhere in the contract — often in the financing provisions — sits a clause that ties the incentive to a determination about whether it threatens the bonds’ tax-exempt status. That clause carries no percentage, no threshold, no number. Just a contingency.
Tax-exempt bond financing limits how operator pay can be structured. So the incentive can be made contingent: it takes one form if the structure is found not to jeopardize the tax exemption, another form if it isn’t. Whether a given structure clears is a question for bond and tax counsel. The financial work is different — price each path, model the incentive’s NPV under the contingency, and identify which fork the operator is facing.
The contingency creates two paths. Path A: the favorable determination issues, and the full double-gate structure takes effect. Path B: it doesn’t, and a fallback applies — the incentive drops to a single gate, tied only to the margin threshold. A model reading only numbers sees double-gate, unearnable, zero. The contract actually creates a fork, and one branch produces a very different gate. The difference is an incentive worth zero versus one worth real fee income across the term.
Nothing in the fee schedule changes. Same percentages, same thresholds. What changes is which gates apply. In Path A, both gates govern, and the incentive stays unearnable because the Index gate fails. In Path B, only the margin gate governs, and the incentive becomes earnable because the margin gate clears. The clause with no number decides which numbered structure the operator actually faces.
A read that treats the fee schedule as the structure misses this. The contingency decides which version of the structure applies. Miss it, and the NPV is wrong by the full value of the incentive across the term. Not a rounding error. Not a sensitivity case. Wrong by the gap between “no incentive upside” and “material incentive upside on one path.”
There’s a deeper layer — which path actually produces better operator economics depends on which gate clears — and that’s a piece of its own. The point here is simpler. The clause forks the structure, and a model that reads only numbers misses the fork.
Negotiable Blanks: The Number That Isn’t There Yet
Sometimes the most important number in an HMA is a blank line.
Performance tests with negotiable minimums show up across the industry. The contract says GOP margin shall be no less than a figure “to be determined by mutual agreement within 90 days of opening.” The blank carries no number, but it decides whether the test is passable. A model that assumes an industry-typical range — 28% to 34%, depending on property type — can miss that the negotiated floor lands at 36%, which turns a clearable test into a fragile one.
The blank pushes the number to a later negotiation, in a different context. By then, performance is visible, the market has clarified, and the relationship has taken shape. How much leverage the operator has over the final number depends on the deal. An operator who delivered strong early performance has more; an operator who stumbled through ramp-up has less — and may watch the blank get filled at a level that locks in failure. The blank is a residual right of control hiding in plain sight: whoever holds the leverage when it’s filled decides its value (Grossman & Hart, 1986).
The consequence is concrete. A test at 32% clears in scenarios where a test at 36% fails. Four margin points is the gap between clearable and fragile. In NPV terms it’s the weighted cost of cure payments, termination risk, and lost operating flexibility across the term. The blank doesn’t just defer a number. It defers whether the operator can pass the test at all without reworking the operating model.
A rigorous read flags the blank, models the plausible range, and identifies what the operator has to win when the blank gets filled. The operator who bids assuming 32% and learns after signing that the owner expects 36% has already committed to economics the deal can’t carry. The blank looked like a detail. It set the whole test.
Force Majeure and Consecutive-Failure Mechanics: The Second-Order Effect
Some carve-outs have a second-order effect you can’t read off the clause.
Performance tests often need two consecutive years of failure to trigger termination. That gives the operator time to correct before the owner gets an exit. Force majeure provisions pull certain years out of the test — a pandemic, a natural disaster, a government-ordered closure. The question is how the two interact. Does a force-majeure year reset the consecutive-failure count, or just pause it?
If it resets, the operator gets a clean start after the disruption. A failure in Year 3, force majeure in Year 4, a failure in Year 5 — that’s one failure, not two in a row. The risk drops sharply. If it pauses, the count picks up where it left off. The Year 3 failure carries forward, and a Year 5 failure becomes the second consecutive one, triggering termination.
The clause often doesn’t say which. It has to be interpreted. And the economics aren’t a rounding error — Monte Carlo termination risk comes out very differently under reset versus pause. It’s the difference between a test that holds up across disruptions and one that carries forward every pre-disruption failure as live termination exposure.
The force-majeure provision has no number. It’s a list of excluded events. But how it meshes with the consecutive-failure count sets the operator’s termination risk across every disruption scenario. A rigorous read spots the ambiguity, models both interpretations, and flags the need for clarifying language. The operator who learns after a disruption that the owner reads it as pause-not-reset carries termination exposure the bid never priced.
Why Contract-Reading Matters
Financial models are necessary and not sufficient. The provisions that flip economics don’t feed a spreadsheet. A rigorous read covers the whole contract — not just the fee schedule and the performance-test section — to find the provisions that decide which structure applies, price each path they create, and flag the ones whose second-order effects need interpretation.
Analysis and counsel run in separate lanes. The analyst prices each path. Counsel decides which path the language actually creates, and whether it holds. The handoff is clean: here’s what each path costs; which one does the contract create? For a hospitality attorney weighing the work, that division is the point — it shows how the financial analysis and the legal read fit together. The analyst doesn’t predict the tax ruling or call how a court reads force majeure. The analyst prices each interpretation and flags which structural questions actually move the money.
A bid that misses these provisions misprices the deal. A contingency or blank discovered after signing can’t be undone. Reading carefully at bid stage costs almost nothing next to getting the structure wrong — because the ambiguity isn’t free. It turns into cost later: the renegotiation and hold-up that transaction-cost economics has long warned about, paid at the moment the operator has the least leverage (Williamson, 1985).
The no-number clauses don’t announce themselves. They sit in financing provisions, general conditions, definitions, carve-out language. They decide which numbered structure applies without holding a number. A model built to pull parameters reads past them. A rigorous read reads for structure.
The Most Expensive Number
The clause with no number changes everything, because it decides which numbered structure actually applies. Models pull numbers. Contracts create structures. A rigorous read needs both.
The most expensive number in an HMA is the one that isn’t there.
Most operators find that out after signing — when the contingency triggers, or the blank gets filled, or the carve-out’s knock-on effect surfaces. The alternative is to read for structure at bid stage, while the economics can still be modeled and the terms can still be moved. The provision that decided whether the incentive was worth millions or nothing didn’t hide. It sat in the contract, governing the structure, waiting for someone to read it.
So on your last bid, how much of the analysis rested on the numbers in the fee schedule — and how much on the clauses that decide which fee schedule actually applies?
Dash Decisions provides vendor-side financial deal desk services for hotel operators bidding on HMAs.