A property reports 96% physical occupancy. Its current-period cash receipts equal roughly 83% of gross potential rent. Both numbers can be true at once. The question worth asking is not which figure is wrong, but what happened between the lease being signed and the money landing in the operating account.
Occupancy Is Not One Number
Physical, leased, economic, and collected occupancy answer different questions, and using them interchangeably is where underwriting goes soft. Physical occupancy is units occupied divided by total rentable units. Leased occupancy counts executed leases, which can include notice units and residents not yet paying. Economic occupancy takes gross potential rent and deducts vacancy, model and administrative units, and bad debt. Collected occupancy, an Emeth analytical measure rather than an industry standard, ties current-period residential cash directly to gross potential rent. Stabilized occupancy is a forward judgment about what the property can sustain after lease-up, turnover, and concessions settle out. Freddie Mac's own guidance requires appraisers to support the stability of rents and occupancy rather than accept a snapshot, which is the same discipline this framework applies at the underwriting desk.
Follow the Revenue From Market Rent to the Bank
Revenue quality is a sequence, not a single line item: market rent, contract rent, billed rent, collectible rent, cash received, cash deposited. Each step can lose ground. Loss-to-lease measures the gap between contract rent and current market rent. Concessions and payment plans reduce what is actually billed. Bad debt and write-offs reduce what is collectible. Security deposits and prepaid rent show up as cash but are not current-period revenue, and under ASC Topic 842, lease income is generally recognized over the lease term while collectibility can limit what gets recognized at all, a distinction that belongs to the property's CPA to finalize. A 200-unit property at $1,600 average market rent produces $320,000 in gross potential rent. After 8 vacant units, loss-to-lease, concessions, and bad debt, effective residential revenue lands at $281,300. After current-period unpaid rent, cash receipts fall to $266,900. That is a 16.6-point gap between physical occupancy and what the bank statement shows for the month, and it did not come from one place. It came from five.
Reconcile the Rent Roll, General Ledger, and Cash
No single report proves anything on its own. Reliability comes from tying independent records together. Start at the unit level: rent-roll charges against tenant ledgers and the monthly charge register. Reconcile total billed rent and concessions to the general ledger's rental-income accounts. Tie resident payments to the receipts journal and the bank deposit batches. Compare the result to the T3, T6, and T12. Fannie Mae's own collection-validation standard requires a cash ledger, a receipts journal, or at least three months of bank statements before rent collections are accepted at face value, and that standard is the right bar for any deal Emeth underwrites, not just agency-financed ones. A rent roll that has not been walked through this chain is a set of scheduled charges, not a revenue statement.
Decide Whether the Gap Is Timing, Execution, or Market Weakness
Every discrepancy is not fraud, and none should be treated as permanent impairment until it has been classified. Timing issues include deposit lag, a system conversion, or a casualty disruption that resolves on its own. Execution problems include weak screening, slow eviction filing, uncleaned ledgers, or concessions that have become structural rather than a lease-up tool. Structural weakness shows up as aging delinquency, repeated write-offs, falling renewal conversion, and renovated units leasing below the underwritten premium. National advertised multifamily rent growth ran 0.2% year over year in June 2026, and Yardi Matrix reported Austin and Tampa among the Sun Belt markets posting annual asking-rent declines, 4.0% in Austin and 2.8% in Tampa. That context does not condemn any single property. It does mean loss-to-lease assumptions built on asking rents in those metros need direct, executed-lease support before they go into a pro forma.
Applying the sensitivity test matters here. If half of a month's delinquency collects within 30 days, normalized monthly cash rises by the recovered amount and the annualized effect compounds. A single month-end delinquency report without the following month's collection history is not evidence of a permanent revenue loss. It is half a data point.
Underwrite the Revenue That Can Be Repeated
The forward revenue line should reflect what the property can earn and collect on a repeating basis, not the best available asking rent or a single strong month. Renovation premiums get validated against executed leases and the payment history that follows, not against the unit's asking price. Recurring concessions get amortized across the lease term rather than treated as one-time. Loss-to-lease assumptions drawn from unleased asking units, rather than signed and paying residents, get treated with real skepticism. Comparing T1, T3, T6, and T12 side by side shows whether concessions and bad debt are improving or getting worse, which a single trailing period cannot tell you. If a full reconciliation shows the $397,200 annual gap between occupied contract rent and cash in the example above is recurring rather than temporary, the value impact ranges from roughly $6.36 million at a 6.25% cap rate to $7.57 million at 5.25%. That is not a rounding error in a model. It is the entire spread between a deal that clears underwriting and one that does not.
A clean reconciliation does not prove a property will perform. It proves the starting revenue is understood, which is the only place any underwriting should begin. A high occupancy number should not receive valuation credit until the residents, leases, concessions, ledgers, general ledger, and bank deposits agree with each other. The working presumption at Emeth is not that a seller has manipulated anything. Property management systems carry timing differences, classification quirks, and ordinary operational noise as a matter of course. The obligation is to explain those differences before a single dollar of assumed revenue gets capitalized into a purchase price.