To: Fox (internal) · Re: 2) Press_House_Proforma_Model.xlsx before it goes to a lender
Asset: Press House at Union District — 331 N St NE, Washington DC (NoMa/Union Market) · 356 units · $110M · 24-month value-add lease-up · Buyer/sponsor Fox, seller Foulger-Pratt JV
Bottom line: The deal is real and potentially attractive, but the model's headline 61% IRR is inflated by a stack of optimistic/erroneous inputs. On corrected, market-grounded assumptions the project return is more like ~33–41% IRR (base) and ~12% (stress). Two things must be resolved before this goes to financing: (1) the capital stack is probably not financeable at $88M / 80% LTV as drawn, and (2) the lease-up/concession assumptions are well below the current NoMa market. Neither is fatal — but the model as written will not survive a lender's underwriting.
Fox is buying a new (2021), Class-A, 356-unit building near Union Market for $110M (~$309K/door). It's only ~87% physically occupied but ~65% economically occupied — a big chunk of "occupied" units aren't really paying (concessions, bad debt, non-revenue units). The plan: over 24 months, fill it up, clean up the rent roll, cut operating costs, lease the empty retail, then sell at a 5.0% cap for ~$164M and roughly 2.7x the equity. The whole return is a bet that the turnaround works — there's almost no margin in the purchase price itself.
The model says 61% IRR / 2.72x. Here's what happens as each issue is corrected (independent recompute, like-for-like):
| Case | Project IRR | Equity multiple |
|---|---|---|
| Model as-is | 61.1% | 2.72x |
| Fix the loss-to-lease bug | 58.4% | — |
| Normalize property tax to run-rate | 52.2% | — |
| Exit at 5.25% / 5.50% / 6.00% cap | 52.8% / 44.7% / 29.3% | — |
| If buyer also pays 2.9% transfer (both sides) | 56.7% | — |
| Lender base case (loss-to-lease + tax run-rate, exit 5.25%) | ~41% | 2.2x |
| Lender base+ (same, exit 5.50%) | ~33% | 2.0x |
| + if buyer also pays full 2.9% transfer | ~30% | 1.8x |
| Stress (slower lease-up + exit 6.0%) | ~12% | 1.4x |
Takeaway: even corrected, this can be a strong value-add return (~33–41%) if the lease-up actually happens. The downside isn't a loss — it's a mediocre ~12% in a bad scenario. The risk is concentrated in execution (lease-up in a soft market) and the exit (cap rate + buyer pool), not in losing money.
🔴 1. The loan probably isn't financeable as drawn. At $88M debt and 6.85% interest, annual debt service is ~$6.0M, but current income (~$4.1–4.6M) doesn't cover it — DSCR ~0.7–0.8x (needs ≥1.0, ideally 1.25x) and debt yield ~4.7–5.2% (bridge lenders want ~6–7%+). A 65%-occupied asset can't get cheap agency debt; it needs a bridge loan, which today runs 65–75% LTV (not 80%) and requires a funded interest reserve — which the model doesn't include. Expect a smaller loan and more day-1 equity than the $26.3M shown. Resolve this first — get the term sheet before approaching a lender.
🔴 2. Lease-up and concessions are priced for a market that doesn't exist right now. The model assumes vacancy falls to 4%, concessions to 1.5% of rent, and 3% rent growth. But NoMa is running ~10–11% vacancy, landlords are giving 2–4 months free, and DC rents fell ~2.2% in 2025. Press House itself currently advertises ~2 months free at ~20% below the area average. Reaching the stabilized income the whole deal depends on, through this market, is the central risk. Re-underwrite lease-up speed, real free-rent, and rent growth to current data.
🟠 3. The price gives no credit for the building being unstabilized. A nearly identical, fully stabilized building two blocks away (The Belgard, same sponsor) sold for $311K/door in 2024. Fox is paying $309K/door for a 65%-occupancy building — essentially stabilized pricing for un-stabilized performance. On trailing income the going-in cap is ~3.8% (the model's 4.15% uses a single flattering month).
🟠 4. A real modeling bug. Loss-to-lease (the gap between market and in-place rents) is in the Projection but dropped from the monthly cash flow that drives the IRR → interim income overstated ~$1.4M, IRR inflated ~3 pts. Restore it.
🟡 5. "Welcome Tax" — mislabeled, but the rate is defensible (you were right to flag it). The line is named with Montreal terminology ("Welcome Tax"), but in DC this is the deed recordation tax, and 1.45% is the buyer's correct recordation rate — the seller customarily pays the separate 1.45% transfer tax (combined 2.9%, but split). So as a buyer cost, 1.45% is fine — it is only understated (~$1.6M, → 2.9%) if Fox agreed to pay both sides in the PSA. Action: rename the line and confirm the allocation in the PSA; no number change needed under standard custom.
🟡 6. Operating-cost cuts look double-counted. The model cuts payroll and cuts contract services 75% "by moving janitorial in-house" — you can't bank both. The resulting ~36% expense ratio is aggressive for a 356-unit building (40–50% typical). A lender's appraiser will add these back.
Comparable / market rents (NoMa-Union Market, 2026)
| Source | Studio | 1BR | 2BR | Note |
|---|---|---|---|---|
| Zumper (H St-NoMa, Mar-2026 median) | $1,729 | $2,355 | $3,013 | submarket medians |
| Apartments.com (NoMa avg) | $1,842 | $2,239 | $3,184 | submarket averages |
| Press House in-place (rent roll) | $2,061 | $2,341 | $3,382 (2BR/2BA) | Class-A premium |
| Press House model target | ~$2,287 | ~$2,705 | ~$3,600 | top of range |
Area demographics (demand backdrop — supportive)
| Geography | Population | Median age | Median HH income | Renters |
|---|---|---|---|---|
| ZIP 20002 | 69,422 | 33.0 | $120,337 (2024) | — |
| Near Northeast | 68,201 | 33.5 | $98,391 | 61.6% |
| NoMa neighborhood | 42,816 | 31.1 | — | renter-heavy |
Gallaudet University (1 block away) — Fall 2025 enrollment 1,260 (807 undergrad / 453 grad), flat YoY (−3), and down over the decade. A stable institutional anchor/employer, but small and not growing → a modest demand support, not a catalyst for lease-up.
Primary source documents ("15 - Press House Dataroom") were tied out to the model. Full audit: ../05_dataroom/DATAROOM_AUDIT.html. The model is faithfully built off the seller's actuals, with these corrections:
Confirmed / favorable - Price $110M confirmed (LOI). Transfer tax: the LOI says the SELLER pays ALL transfer + recordation, so the buyer pays $0 - the "Welcome Tax" $1.595M is a ~$1.6M conservative OVERSTATEMENT (remove or keep as cushion). - Property tax validated: ~$1.12M main lot, FLAT 2023-2026, no reassessment. - Equity is institutional (Farallon, $42B AUM, no fundraising contingency). - Retail "+$35K/mo" upside is largely CONTRACTED via executed Starbucks ($122,720/yr) + Scissors & Scotch leases (rent-commencement timing risk, not leasing risk).
Confirmed concerns (now quantified) - Going-in NOI cherry-picks April ($360,785/mo) vs the T12 average ($324,464/mo) - entry ~11% high; normalized going-in cap ~3.6%. - Concessions are rising, not falling: Jan $11K -> Apr $125K/mo (~12.6% of GPR) vs the modeled 1.5%. - Collections: AR $589K, 108/356 units (30%) delinquent, ~$227K in 60-90+/over-90 - the April bad debt (-$8.87) is not credible. - Capex: $977K/yr "non-recurring" actual vs the $70K FFE line. - Still open: the $88M senior loan (Farallon is equity, not the lender).
Per Fox's direction, two assumptions were corrected in the REVIEWED workbook and the investment deck: - Concessions: held at the current ~12.6% of GPR (April actual ~$125K/mo) for 12 months, then trending down (4% by month 24, 3% stabilized) — vs the original glide to 1.5%. - Other income: flat for the first 6 months, then growing at an annual rate ramping 3% -> 6% (months 7-24).
Corrected model result: Project IRR 53.9% / 2.42x equity multiple (vs the 61% headline); net equity at exit ~$69.0M; stabilized NOI ~$8.0M; gross exit value ~$160.6M; yield on cost 6.8%. The investment deck has been updated to these figures.