A REPE model is the set of spreadsheets and assumptions that turn a property business plan into projected cash available for distribution and investor returns. A “lender definition” is the bank’s contract language for what counts as cash flow for sizing, covenants, sweeps, and extensions. Most return damage comes when the model uses the sponsor’s definitions while the deal runs on the lender’s.
REPE models rarely break because someone can’t do math. They break because the model draws the boundary in the wrong place. Outputs still look reasonable at underwriting, then show up later as a refinancing gap, a covenant trip, tax cash needs, or distributions that arrive later than promised. IRR doesn’t forgive timing.
A useful rule is simple. If a line item affects (i) debt proceeds or covenants, (ii) tax basis and taxable income timing, or (iii) the partnership waterfall, treat it as a gating assumption. Give it an owner. Tie it to a document. If it’s “owned by the model,” it’s usually owned by nobody.
Why “definition risk” is the silent killer in real estate models
Definition risk is the gap between how your spreadsheet labels cash flow and how your legal documents define it. That gap matters because lenders and partners are not paying you based on your tabs and formulas. Instead, they pay you based on defined terms in the credit agreement, the JV agreement, and the closing statement.
To make the payoff concrete, think of the model as a map and the deal documents as the terrain. When they disagree, you do not get “close enough.” You get surprises that show up as blocked distributions, unexpected equity calls, or an extension you assumed you could exercise but cannot.
1) Size debt using the lender’s cash flow, not sponsor NOI
Most drift starts with a simple mistake. The sponsor underwrites and sizes debt on “NOI,” but the lender sizes and tests on a tighter definition. The model then shows higher proceeds, less equity, and earlier distributions. The credit agreement delivers something else: lower proceeds, more reserves, and tighter cash control.
Start with what lenders actually do. For stabilized multifamily, sizing tends to hinge on underwritten NOI, DSCR, and LTV. For office, retail, and industrial, the definitions vary, and the exclusions matter. The cash flow in your model is not “NOI.” It is “NOI as defined in Section X of the loan.”
Common mismatches that move real money
- Bad debt and concessions: Sponsor models often gross up rent and net down later, while lenders haircut gross potential rent and treat concessions as a steady drag until collections prove otherwise.
- Management fee floors: If you self-manage, you may model an internal cost, but many loans require a fixed percentage of effective gross income whether you pay it out or not.
- Reserves and escrows: Replacement reserves and TI/LC escrows are sometimes treated as operating expenses for DSCR, while sponsors often place them below NOI as capex.
- Free rent and downtime: Commercial lender NOI often removes speculative steps, adds downtime, and bakes in leasing costs, which shrinks proceeds and tightens tests.
The damage is not limited to proceeds. Trigger behavior changes. Cash sweeps, lockbox hardening, and distribution blocks activate on the lender’s DSCR and debt yield, not your “sponsor DSCR.” As a result, operations can look fine while the lender definition fails because of reserves, fees, or timing.
Practical discipline is straightforward. Build a lender NOI bridge that reconciles sponsor underwriting to the credit agreement line by line. Tie each adjustment to a defined term in the term sheet or draft. Then model both sizing and ongoing tests on the lender definition, even if the documents aren’t final. Use a conservative market definition and sensitize the key exclusions.
Reserves under lender control are not “just accounting.” If the cash sits in a controlled account, it is not available to pay investors. The model should treat it that way.
Kill test: If the deal requires DSCR within 0.05x of the minimum or LTV within 1% of the cap to hit your proceeds, the lender is effectively your investment committee. Price the equity for that reality.
2) Treat the exit as a financing problem, not just an exit cap rate
Most models treat the exit as a valuation exercise: apply an exit cap rate to year-N NOI and subtract debt. In the real world, the exit is a buyer underwriting and financing feasibility problem. You can be “right” on cap rate and still be wrong on proceeds.
First, terminal value often ignores the quiet deductions: escrows, holdbacks, working capital adjustments, property tax prorations, tenant claim reserves, and lender release costs. Brokerage and transfer taxes show up more often; the rest get waved away. On a single-asset deal, those items can move meaningful basis points of IRR because they hit at the finish line.
Second, many exits are de facto refinancings. Even if you sell to another sponsor, that buyer typically uses leverage. If the buyer’s debt sizing is constrained by DSCR and the rate environment, your price is constrained. A model that assumes a “cap rate sale” without a buyer debt stack is assuming an all-cash buyer or a different market.
A tighter exit framework with three linked blocks
- Buyer underwriting: Start with stabilized NOI after buyer-required reserves, management fees, and any capex needed to reach “lender stabilized.”
- Buyer financing: Size debt off plausible DSCR, interest rate, amortization, and product constraints (bank, debt fund, CMBS).
- Equity check: Reconcile the implied purchase price to what the financing block supports, and haircut the exit or move work into pre-sale capex if it does not clear.
Kill test: If your base case shows strong terminal value growth but a plausible refinance would not pay off the remaining loan balance, your “exit” is not a plan. It’s a bet on conditions you don’t control.
3) Model capex timing and controls like governed cash, not a suggestion
Capex mistakes are rarely about the total budget. They are about timing, escrows, approval mechanics, and who holds the keys to the cash. Because value-add and repositioning plans are schedule-driven, timing errors often become liquidity problems before they become valuation problems.
Many facilities require capex and TI/LC escrows. Draws can require invoices, lien waivers, inspections, and lender sign-off. The money may be “yours,” but it is not always usable when you want it. Models that assume capex can be delayed to protect distributions often collide with loan terms or the business plan.
Three capex misses that show up in cash
- Interest carry and fees: Value-add work often depresses NOI before it lifts it, while interest, lender fees, and reserves can exceed the model’s “capex” line.
- Expense classification: Some building costs hit the P&L depending on policy and lender definitions, which changes DSCR and even sizing.
- Trapped operating cash: Lockbox and sweep provisions can push cash into controlled accounts when covenants tighten, blocking capex unless you inject equity.
A decision-useful model treats capex like a small project finance structure. Separate sources and uses by funding type: operating cash, sponsor equity, capex reserve, future funding, preferred equity. Add draw lags and holdbacks. Most important, tie rent steps and occupancy gains to capex completion. Don’t let rent growth appear unless the enabling work is funded and spent.
Kill test: If the model pays distributions while a lender-controlled reserve is below its required level, those distributions may be blocked. That is the contract.
4) Drive the waterfall and promote from documents, not templates
Promote economics look clean in a spreadsheet. They look less clean in a partnership agreement once you include fee offsets, recycling, tax distributions, catch-ups, and clawbacks. Because the waterfall defines who gets paid and when, small drafting choices move real money between LPs and the sponsor.
Model the boundary conditions from the JV agreement or LPA, not from memory. Then confirm the timing convention, because that is where many “we’re in promote” misunderstandings come from.
Boundary conditions that change economics
- Capital definitions: Confirm what counts as contributed capital and return of capital, including fees, closing costs, lender fees, and working capital.
- Preferred return mechanics: Check simple versus compounded, monthly versus annual, and what base it accrues on (unreturned capital only or also unpaid pref).
- Recycling and catch-up: Confirm reinvestment rules, caps, consent rights, and the exact GP catch-up language.
- Clawback and tax distributions: Treat clawback as real if backed by escrow or guarantee, and model tax distributions that compete with pref tiers.
Then separate the cash waterfall from tax allocation assumptions. Under US partnership rules, taxable income can diverge from distributions. If the model assumes “cash equals taxable,” it will misstate tax distributions and after-tax returns.
Kill test: If a small change in distribution timing flips the deal “into promote,” confirm the agreement’s timing convention. Some test at each distribution date; others test at liquidation.
5) Put fees and transaction costs in the right period (and in the right bucket)
REPE returns are sensitive to timing. A dollar that leaves at close hurts IRR more than a dollar that leaves in year five. Fee errors are usually not subtle. They are missing items, wrong payer, wrong period, or wrong treatment in the waterfall.
Fee items that commonly get mis-modeled
- Debt fees and OID: Arrangement fees, lender legal, third-party reports, and OID often net from proceeds or require cash at close, increasing the equity check.
- Hedging liquidity: Caps require a premium, and swaps can require collateral under a CSA even if it is expected back later.
- Tax and insurance resets: Reassessments and premium step-ups often hit soon after close, not smoothly over time.
- Disposition and prepayment: Yield maintenance, defeasance, or exit fees can reduce sale proceeds materially.
Leakage is also about where fees sit relative to the waterfall. An acquisition fee paid from equity is different from a fee offset against promote or treated as a distribution. The model must match the documents and side letters.
Controls help. Maintain a fee schedule with payer, timing, whether financed, whether capitalized, and whether offset in the waterfall. Reconcile closing uses to the settlement statement and loan closing statement. If you can’t tie within a tight tolerance, the IRR is a guess.
Kill test: If the deal only works assuming no prepayment cost, confirm the debt product. Many fixed-rate structures make that assumption unrealistic.
6) Model floating-rate debt as a contract with triggers, not “SOFR + spread”
Floating-rate exposure is not a single rate. It is a contract with conventions: floors, lookbacks, compounding, payment frequency, margin step-ups, default interest, extension tests, and cash management triggers. Modeling it as a simple periodic interest line can understate cash interest and hide breakpoints.
Rate floors matter, and so do compounding and lookback conventions. IO can end early if milestones are missed. Extensions can require both fees and meeting DSCR, LTV, or debt yield tests. Sweeps and distribution blocks do not change NOI, but they change what reaches investors, which is why timing damage can compound quickly.
Hedging belongs inside the debt model. If a cap is required, model the premium at close, the strike, the notional schedule, and the term versus loan maturity and extensions. Also model counterparty constraints and replacement risk if the loan requires approved counterparties.
Kill test: If base-case DSCR looks healthy but the deal still shows tight liquidity after capex, assume you missed a reserve, a sweep, or an extension fee. Liquidity events arrive before valuation events.
7) Don’t let structure and tax become “footnotes” that trap cash
Sponsors often treat structure as footnotes. In complex deals, structure drives cash. Lenders require SPE provisions, separateness covenants, independent directors, and restrictions on upstreaming cash. If your model assumes cash moves freely among entities, check reality.
This breaks in multi-property structures, cross-collateral limits, property-level preferred equity with its own lock, mezzanine debt with intercreditor payment blocks, or lender-controlled accounts with periodic releases. In each case, cash can exist but not move to where it’s needed, which changes both stabilization risk and distribution timing.
Tax and withholding also cannot be hand-waved. Withholding on cross-border distributions, state and local transfer taxes, and investor-required tax distributions are cash items. They reduce net returns and can change pricing and partner economics. If you have cross-border complexity, align your placeholders with counsel and consider a dedicated memo, even if the underwriting model stays high level.
Operating discipline: make the underwriting model the operating model
Avoiding these errors is mostly ownership and documentation. At term sheet, build a definitions schedule for lender NOI, covenants, reserves, extension tests, and cash management triggers. At draft-doc stage, update the model to match the credit agreement, JV/LPA waterfall, and fee schedule from engagement letters and closing estimates.
Pre-close, tie the model to the settlement statement and loan closing statement, and confirm hedging premium and counterparty documents. Post-close, report and monitor using the lender’s definitions monthly, because that’s what governs sweeps and extensions. If you need help formalizing this process, it can also be useful to standardize your capital stack assumptions and the way you schedule debt and reserves, so operating teams and finance teams are reading the same playbook.
A fresh angle: version control is a return driver, not just “good hygiene”
Model risk is not only about assumptions. It is also about provenance. When a covenant trips or a refinance gap appears, the first operational question is often, “Which model did we actually run, and what changed?” If you cannot answer that quickly, you will lose time, negotiating leverage, and sometimes money.
Archive the final model and all supporting files with an index, version history, Q&A log, named users, and full audit logs. Hash the archived package and store the hash with the deal record. Apply a written retention schedule, then require vendor deletion with a destruction certificate when retention ends. Legal holds override deletion.
Fast screens help. Reconcile gross loan to net cash after OID, fees, reserves, and hedging. Build a clear lender NOI bridge. Run a mild downside to see whether sweeps block distributions. Model extension conditions and fees. Drive the waterfall from documents. Finally, test the exit under buyer debt terms, not just cap rates. This is where structured stress testing can complement traditional sensitivity work; for more on that discipline, see stress testing approaches used in REPE underwriting.
Key Takeaway
A REPE model becomes reliable when it stops being a sales document and starts being a document-driven operating tool. If you bridge lender NOI, govern capex cash, model the waterfall from the agreement, and treat exits as financeable transactions, you reduce the exact timing shocks that IRR punishes most.