The series valued four HMA options one at a time — the performance test, cure, the termination fee, key money recovery. Each piece showed the option’s structure and what it’s worth in isolation. The performance test is the owner’s termination option, triggered by failure and exercised along several paths. Cure is the operator’s compound option to extinguish that termination right. The termination fee is the strike price on owner-initiated termination. Key money recovery is the barrier option that governs what the operator collects when termination happens.
But an HMA isn’t a list of independent options. It’s one nested system where the same events move several at once.
An HMA is a system of nested, correlated options: the same events move several at once, so its true value is less than the sum of the options valued separately — the non-additive interaction of options written on the same underlying (Trigeorgis, 1996; Dixit & Pindyck, 1994).
Each of these options is legitimate in its own right — the owner’s termination remedy, the fee it pays to exercise, the forfeiture and settlement rules — as the component pieces set out. This piece is about how they interact, and it reads that interaction from the operator’s side.
Value them separately and the operator double-counts protection that’s really one correlated bet. Cure, the termination fee, and key money recovery all look like independent protections when valued one at a time. But in the underperformance state — the state where the operator needs protection most — all three fail together. They’re one correlated bet, not three separate hedges. So the operator mis-prices its true exposure.
What follows: the nesting — how the four options chain into one structure; the correlation — how shared events move several at once; a worked example showing the gap between sum-of-parts and system value; and what the operator does with it.
The Nesting: How the Four Options Chain
The four aren’t parallel options. They’re nested — one sits inside another, and exercising or extinguishing one changes what the others are worth.
The performance test is the trigger layer. Failure creates the owner’s termination option. “Performance Tests as Option Contracts” set out this structure; “What a Performance Test Is Actually Worth” valued it as the owner’s expected payoff across the exercise paths — terminate, cure payment, renegotiate, do nothing. The test is the master trigger; everything downstream depends on whether it fires.
Cure sits inside the termination option. The owner can terminate only if cure isn’t or can’t be exercised. “Cure Rights Are a Compound Option” showed cure as the option nested in the termination option — the operator pays or performs to extinguish the owner’s right. Cure is the first line of defense: if it succeeds, the termination branch never opens; if it fails, the owner holds the decision.
The termination fee is the strike on owner-initiated termination. “The Termination Fee Is a Strike Price” showed how the LD fee prices how often the owner exercises the convenience or sale branch. A high fee deters exercise; a low or zero fee (for-cause) makes it cheap. The strike governs which termination path the owner takes.
Key money recovery is the settlement layer. “Key Money Recovery Is a Barrier Option” showed what the operator collects or forfeits when termination happens, gated by two barriers — termination type (for-cause forfeits) and asset value (recovery only if the hotel is worth more than its debt). It’s the final settlement: what the operator walks away with after the decision is made.
The skeleton runs trigger, then cure, then the termination branch (for-cause versus convenience/sale), then settlement. Not four separate options — one connected structure where each link affects the next.
Nesting shows the structure. It doesn’t yet explain why the system is worth less than the sum of its parts. The correlation does.
The Correlation: The Same Events Move Several Options at Once
The options aren’t independent, because the same underlying events move several at once. In the bad states, the protections fail together.
Underperformance is the master driver. One bad-performance state does four things simultaneously: it pushes the performance-test option toward exercise (the test fails, the owner holds the termination right); it makes cure harder or impossible (thin margins, no operational room, an unaffordable cure payment); it drives for-cause termination, which zeros the LD strike (no fee on for-cause); and it knocks out key money recovery on both barriers (for-cause forfeiture on the type barrier, depressed value on the asset barrier). In the underperformance state, cure, LD, and key money recovery all fail at once. One correlated bet, not three independent protections. An operator who values each separately and adds them up over-counts protection in the state where the operator needs it most.
Sale is the second shared event — correlated in the benign direction. A sale fires the convenience/sale termination right (the LD fee pays) and key money recovery (healthy asset, no fault, both barriers clear). Here the protections pay together, in the operator’s favor: it collects the fee and recovers the key money in the same event.
Benign drift is the third state. Performance holds, no sale, the contract runs. Nothing exercises. The base case where the system just operates.
The valuation error is treating each option under its own marginal probability — cure under “how often is cure needed,” LD under “how often is there a sale,” key money under “how often is there a termination” — as if they were independent. They aren’t. The underperformance probability drives all of them. You can’t count cure, LD, and key money as three separate hedges when they all vanish in the same state.
The gap is material. A worked example shows how much.
One Worked Example: The Double-Counting Gap
Take one illustrative deal. The operator faces termination exposure from the performance test, and three protections offset it: cure, the termination fee, and key money recovery. Value the three two ways — sum-of-parts (naive) and system (shared scenarios).
Start with the states. Three of them, with per-state payoffs:
- Underperformance (35% probability). The test fails, for-cause termination is likely. Cure fails (thin margins). The LD fee pays nothing (for-cause carries no fee). Key money recovers nothing (both barriers fail). All three protections pay zero.
- Sale (20% probability). The owner exercises a convenience or sale termination. Cure isn’t needed. The LD fee pays $4.2M. Key money recovers the full $5.8M (no fault, healthy asset). Total protection in this state: $10.0M.
- Benign drift (45% probability). No termination, contract runs. Nothing triggers, nothing pays.
The naive approach — sum of parts. Value each protection under its own assumed probability, and each gets over-weighted.
Cure: an 18% assumed probability against a $2.4M cure payment avoided — about $0.43M. It ignores that cure fails in the underperformance state, exactly where it’s most needed. LD: a 28% assumed probability against the $4.2M fee — about $1.18M. It ignores that for-cause termination, the dominant mode in underperformance, carries no fee. Key money: a 32% assumed probability against the $5.8M balance — about $1.86M. It ignores that most early termination is for-cause, where the type barrier knocks out recovery, and happens in underperformance, where the asset barrier fails too.
Add them: about $3.47M. The naive approach over-weights the good state (sale) and under-weights the correlation that fails all three together in the bad state.
The system approach — shared scenarios. Value the protection total in each state, then weight by probability. Underperformance contributes nothing (all three pay zero). Benign drift contributes nothing (nothing triggers). Only the sale state pays, and it pays $10.0M — but only 20% of the time. Twenty percent of $10.0M is $2.0M. System value: $2.0M.
The gap. Sum-of-parts says $3.47M. The system says $2.0M. The gap is $1.47M, and every dollar of it is the correlation.
Look at the probabilities. The naive approach put 78% of probability mass on protection payoffs — 18% for cure, 28% for LD, 32% for key money. The system approach puts 20% — only the sale state. The 58-point gap is the correlation: the naive approach counted protection in states where the protections don’t actually pay, in underperformance (where all three hit zero together) and benign drift (where none are needed). Broken down, the over-count is about $0.43M on cure (an 18% assumed probability on a protection that pays 0% of the time), about $0.33M on LD (28% assumed against 20% actual — the extra points are for-cause terminations with no fee), and about $0.70M on key money (32% assumed against 20% actual — the extra points are underperformance for-cause terminations where both barriers fail).
The three protections aren’t three hedges. They’re one correlated bet that fails together in the 35%-probability underperformance state and pays together only in the 20%-probability sale state.
What the Operator Does With It: System-Level Strategy
The remedies aren’t per-option tweaks. They’re system-level moves that change how the options correlate.
Value the system under shared scenarios, not each option under its own probability. The operator’s true protection value is the scenario-weighted system total, not the sum of the per-option values. Identify the shared events — underperformance, sale, benign drift — value the protection total in each, weight by probability, and add. This is transparent expected-value work, not a closed-form multi-option formula. It shows where the protections actually pay and where they fail together.
Diversify the correlation, not just the options. Because cure, LD, and key money all fail in the underperformance state, adding a fourth protection that also fails there buys nothing. The valuable protection is one that pays in the bad state — uncorrelated with the master driver. Key money made refundable on for-cause termination breaks the type barrier and decouples key money from the underperformance driver, so it pays even in the bad state. A performance-test carve-out for owner-caused failure breaks the trigger correlation — if the owner’s own actions drive underperformance, the test doesn’t fire, so the three protections don’t collapse together. A cure payment cap that’s affordable even in thin-margin states keeps cure alive in the underperformance state instead of going to zero. Spend negotiating capital on protections that don’t all vanish in the same state.
Find the single highest-leverage term — the one that breaks the most correlation. One term can improve the whole system more than three independent tweaks. Making key money refundable regardless of cause both moves the key money barrier and partially decouples the settlement layer from the underperformance driver. In the example above, that one change lifts the underperformance-state key money recovery from zero to $5.8M — the balance now pays even on a for-cause exit — which lifts the system protection total from $2.0M to $4.0M. A $2.0M improvement from a single term. That’s system-level leverage, not a per-option tweak. The highest-leverage terms are the ones that pay in the underperformance state, because that’s where the naive approach over-counted and where the system is most exposed. A term that moves protection value from the 20%-probability sale state to the 35%-probability underperformance state is worth more than its face value suggests — it’s breaking the correlation that makes the system fragile.
The lane handoff. Enforceability, drafting precision, and regulatory constraints are counsel’s lane. The operator’s job is to see the system, value it under shared scenarios, and target the correlation-breaking terms. The financial analysis finds which terms carry system-level leverage; counsel drafts them to hold.
The system view changes what the operator negotiates for. The naive approach says get better cure terms, a higher LD fee, better key money recovery — three separate asks. The system approach says get one term that breaks the correlation — one ask with three times the impact.
The Interactions Are the Entire Story
The individual pieces valued four HMA options separately, and correctly, in isolation — each worth what it’s worth when everything else is held constant. But an HMA doesn’t hold everything else constant. The options nest, one inside another, and they correlate: the same events move several at once.
Value them separately and the operator double-counts protection — treating cure, LD, and key money as three independent hedges when they’re one correlated bet that fails together in the underperformance state. The system is worth less than the sum of its parts, because the protections the operator is counting on vanish in the state where it needs them most.
So value the system under shared scenarios. Diversify the correlation, not just the options. Target the single term that breaks the most correlation. That’s the strategy the sum-of-parts view can’t see.
An HMA isn’t a list of options. It’s one nested, correlated system. So the question for the next deal isn’t what each protection is worth on its own — it’s this: when the operator adds up the cure rights, the termination fee, and the key money recovery, is it counting three hedges, or one bet that pays in a state it may never reach?
Dash Decisions provides vendor-side financial deal desk services for hotel operators bidding on HMAs.