Ground-Up Development Model — Marvell Rock Hill (246-Unit Multifamily), 2114 Riverchase Boulevard, Rock Hill, SC

The model is built and fully populated — but the deal as underwritten doesn't clear. At a market-calibrated cost basis, this project produces a -69 bps development spread (trended) against the Charlotte MSA multifamily market cap rate, a 1.9% levered IRR over the 10-year hold, and zero GP promote — the waterfall never gets the LP past its 8% preferred return. 1

Why it doesn't clear. The site is genuinely well-positioned — zero market-rate apartment supply under construction in Rock Hill, 1,500+ announced jobs at Palmetto Research Park (Novant Health, Octapharma), and comps averaging $1,677/unit on ~984 SF 2 — but at today's construction-cost and cap-rate environment, a $280K/unit all-in basis (hard cost ~$190K/unit + site work ~$18K/unit + soft costs/FF&E/contingency/fee) doesn't generate enough stabilized NOI relative to a 5.15% market cap rate 3. The Supportable Land/Acquisition Basis the model solves for — the land price at which a 150bp spread over market cap clears — is $0 4: even acquiring the land for free wouldn't fix the spread at this cost basis. The gap is on the cost and rent-growth side, not the land price.

What's driving the softness, and what would move it:

• Construction cost is the single biggest lever. Non-land development cost is $66.3M against a supportable ceiling of $46.2M 5 — hard costs would need to fall roughly 30%, or rents/exit value would need to rise by a comparable margin, to close that gap. This is a market-calibrated estimate (Charlotte-area 2026 comps run $150-350/SF for garden multifamily); the Blythe Development site-work bid attached to this deal explicitly excludes the multifamily parcel ("the multi-family will be bid at a later date") 6, so an actual GC bid on the MF buildings is the highest-value next step before this deal is walked away from.
• Rent growth is a genuine open question. The datamart's own trend engine flags low confidence on submarket rent growth — Rock Hill-west in-place rent is down 2.55% trailing 12 months on thin history, a "cyclical drawdown" the engine can't confirm is temporary 7 — against the OM's third-party forecast of +3.8%/yr through 2030 on the strength of zero new supply and the job announcements 8. I used a tempered 3.0%/yr; if the OM's more bullish read holds, the spread gap narrows but doesn't close on its own.
• Break-even rent for zero spread is 111.9% of the underwritten roster 9 — asking rents nearly 12% above the $1,676 blended average calibrated to the OM's own comps. That's a stretch relative to today's comp set, even before construction-cost relief.

Key assumptions behind these figures (flagged where estimated): unit mix (98 1BR/750SF, 111 2BR/1,100SF, 37 3BR/1,300SF) is an estimate calibrated to the OM's rent comps, since no unit-mix breakdown exists in the deal documents; land price ($2.5M) and hold period (10 years) were your inputs; hard-cost escalation (3.5%/yr), stabilized occupancy (95%), and exit/market cap rates (5.40%/5.15%) were confirmed with you against datamart and forecast-engine reads.

The full model — Assumptions, Dev Budget, Draw Schedule, Sources & Uses, Pro Forma, Debt Schedule, Returns Summary, Waterfall, Sensitivity, Break-Even, Rent Comps, and Sales Comps — is live and linked; every check (Sources & Uses balance, budget tie-out, loan sizing convergence, LTC targets, roster/NRA tie-out, waterfall tie-out) reads PASS with zero formula errors. Every input cell is exposed on the Assumptions and Dev Budget tabs so you can flex land price, hard cost, rent growth, or exit cap directly and watch the spread move.

Happy to run a goal-seek on any single lever — e.g., "what hard-cost-per-unit clears a 150bp spread" or "what exit cap makes this work at $190K/unit hard cost" — once you have a real GC number for the multifamily buildings.