Playbook › Playbook
A guided margin-expansion rate is a finite lever quoted as if it were perpetual, and the ceiling never appears in guidance
Claim
[[pattern-buyback-manufactured-eps-screens-as-growth]] catches one half of this: EPS growth that is share count rather than business. This note covers the other half — the margin term — which is subtler, because unlike a buyback it does not need funding and so carries no visible constraint.
A typical algorithm reads: "high-single-digit revenue growth + 100–150bps/yr margin expansion + buybacks = 12–15% EPS CAGR." Every term is presented on the same footing. But they are not the same kind of term:
| Term | Constraint | Visible in any ratio? |
|---|---|---|
| Revenue growth | market size, share | yes |
| Margin expansion | 100% minus the cost floor | no |
| Buyback | FCF and a leverage cap | yes (net debt/EBITDA vs target) |
The margin term is the only one with an invisible ceiling.
The test — extrapolate the guided rate and read the terminal margin
IHG, 2026-08-26: fee margin 65.9% (H1 2026), guided to expand 100–150bps/yr with no stated end. Run it forward at the midpoint:
| Year | Fee margin |
|---|---|
| 2026 | 65.9% |
| 2031 | 72.2% |
| 2036 | 78.5% |
No hotel franchisor has sustained a 78% fee margin. Asked directly when the expansion stops, the CFO said only "we wouldn't put a time limit on it at this point." That is the tell — not a denial, an absence.
What it means for the terminal multiple
Decompose the delivered growth over a full cycle, not the guided growth. IHG FY2019→FY2025:
| Layer | Contribution |
|---|---|
| Rooms / segment revenue | +2.9%/yr |
| Margin expansion (41.5% → 51.3% segment op margin) | +3.6pp |
| Buyback | +2.2pp |
| = Adjusted EPS | +8.7%/yr |
Two-thirds of the EPS growth came from the two finite levers. When they exhaust, the business grows at ~3%. A terminal multiple should be paid against that number, not against the 12–15% headline — which is why a reverse DCF on this shape almost always demands both sustained high growth and no multiple compression.
Guard
- Extrapolate the guided bps/yr for ten years and state the implied terminal margin. If it is absurd, the algorithm has an undisclosed expiry.
- Ask what the margin ceiling actually is — the cost floor of the business, not a round number. For a franchisor it is the cost of running the brand, sales and reservation system.
- Check whether the buyback leans on re-levering. If net debt/EBITDA is guided from the bottom to the top of a target range, part of the buyback is a one-time balance-sheet action wearing the clothes of a recurring return.
- Anchor the terminal multiple on revenue growth, not EPS growth.
What would falsify this
A company sustaining guided margin expansion long enough that the terminal-margin extrapolation stops being absurd — i.e. the expansion decelerates smoothly toward a stated ceiling rather than stopping. Management naming an explicit target margin would convert this from a hidden lever into a disclosed one, and the pattern would not apply.
Related
[[pattern-buyback-manufactured-eps-screens-as-growth]] — the share-count half of the same arithmetic. [[pitfall-growth-ratio-against-collapsing-base]] — the other way a growth rate misleads. [[pitfall-gate-cleared-by-adjective-not-arithmetic]] — "we wouldn't put a time limit on it" is the adjective standing in for the arithmetic.
History
- 2026-08-26 — written off IHG, where the fee-margin term is the load-bearing half of a 12–15% EPS algorithm. Cross-checked against IT and GDDY, where the same shape appears.