How the model is built, and how it is checked
A single deterministic cash-flow model, per unit, EGP nominal, over a 10-year term, run for three structures across three markets and two fit-out tiers. The six tests below all pass.
The three structures
The operator (ZaCoastal) finishes and furnishes a core-and-shell unit and operates it as a rental under 100% management. The owner keeps the unit but has no personal use; the fit-out is repaid in one of three ways:
- O1 — 100% cash upfront. The owner pays the full cost plus markup in cash; the revenue share starts from first operation.
- O2 — partial cash. The owner pays part of the cost plus markup in cash; the rest is recouped from operating income; the revenue share starts once recouped.
- O3 — 100% from operating income. The owner pays nothing upfront; the operator recoups the full fit-out from operating income; the share starts once recouped.
The model walks through three phases: fit-out (capex and overrun risk), recoup (NOI is applied to the operator’s balance first), then split (NOI is shared between operator and owner).
Key parameters (base case)
| Parameter | Symbol | Base value | Meaning |
|---|---|---|---|
| Cash markup | m | 15% | Markup charged on the cash portion of fit-out. |
| Deferral markup | m_def | 25% | Markup on the deferred (recouped) portion; the report argues it must be fixed at signing. |
| Cash share | p | 50% | Fraction of fit-out paid in cash upfront (O2). |
| Cash-markup multiplier | k | 0.5 | In O2 the cash-part markup is m × k. |
| Recoup rate | r | 100% | Share of NOI applied to recoup during the recoup phase. |
| Operator NOI share | s_op | 30% | Share of NOI after recoup (base case). |
| Term | — | 10 yr | Model horizon. |
| Hurdle | — | T-bill + 8pt | Discount rate for operator NPV est. |
| VAT on short stays | — | 0 / 14% | Modelled both ways; unresolved legal question. |
Exact definitions live in za_model_v3.py; this page summarises the six tests, not the source code.
The six tests
All six pass. They are the reason the numbers in the report can be trusted as internally consistent, even though the inputs they run on are still unverified.
| Test | What it verifies | Result |
|---|---|---|
| Ledger conservation | Gross = VAT + opex + refurbishment + operator take + owner receipt, in all options / geographies / VAT settings. | PASS |
| Closed-form recoup | Operator total take = min(10·N, B0) + s·(10·N − B0) with constant NOI. | PASS |
| Limiting cases | Zero share / zero markup loses; huge O1 markup wins; zero revenue → loss. | PASS |
| Monotonicity | NPV up in revenue and share; owner surplus down in share; overrun hits the operator only. | PASS |
| O2 reduces to O1/O3 | O2(p=1, k=1) = O1; O2(p=0) = O3. | PASS |
| Hand calculation | O1 with m=0, s=0, no overhead, no tax → operator NPV = 0. | PASS |
What passing these tests does — and does not — prove
The six tests prove the arithmetic is consistent: every EGP is accounted for, the recoup formula is closed-form correct, the limits behave, and the results move the right way when inputs move. They do not prove the inputs are right. The realised gross yield, fit-out cost and hurdle are still unverified calibrations — and the realised yield is the single largest driver of every result.
Treat this as a model, not a forecast. The full list of unverified inputs and the unresolved legal items is on the assumptions & risks page; the headline findings and complete tables are in the full report.