← Underwriting notes

Loss-to-lease is two numbers, not one

June 12, 2026 · 6 min read

Loss-to-lease is the simplest metric in a rent roll: market rent minus in-place rent, annualized. A 24-unit building with units averaging $150 under market carries $43,200 of LTL. Brokers love the number because it reads as found money — buy the building, mark the rents to market, collect the difference.

The problem is that LTL as a single number answers the wrong question. The right question is not “how far below market are the rents?” It is “which of these dollars can a buyer actually collect, and when?” That split — recoverable versus locked — is a function of three things the headline number ignores.

What determines recoverability

  • Tenure. A tenant who moved in eight months ago at $1,450 against a $1,500 market is functionally at market; the lease rolls in a year and resets. A tenant who moved in twenty years ago at 40% below market is a different asset entirely. Build a tenure histogram from move-in dates: under 3 years (already near-market, limited upside), 3–10 years (moderate turn probability), 10–20 years (low), over 20 years (near-zero). The usual finding is uncomfortable: the LTL dollars concentrate in exactly the buckets least likely to turn.
  • Regulatory regime. In rent-controlled jurisdictions — San Francisco, New York, Los Angeles, Oakland, Berkeley, Santa Monica, Portland, St. Paul, Minneapolis — the LTL on long-tenured units is structurally locked absent vacancy decontrol. Annual increases are capped at CPI-indexed or fixed percentages regardless of where market rent goes.
  • Lease structure. Rolling annual leases reset on a schedule you can model. Multi-year terms and just-cause eviction overlays don't.

Mechanism is not timing

Vacancy decontrol means the mechanism for a rent reset exists when the unit turns. It says nothing about whether the unit turns. A 25-year tenant in a rent-controlled, just-cause jurisdiction is recoverable in mechanism and immortal in practice. Underwriting that unit's LTL as recoverable because “the jurisdiction has vacancy decontrol” conflates a legal right with a cash flow.

A memo we regression-test against once classified $42,048 of LTL as “recoverable / $0 locked” — three sentences before admitting the rent roll disclosed no lease dates. The two claims are incompatible. Without tenure data, the recoverable-locked split is unknowable, and the honest output is “loss-to-lease $42,048, recoverability indeterminate — tenure data required.” That failure is now a pinned test in our pipeline.

What to do with a rent roll that lacks lease dates

  • Report total LTL as one number, labeled “tenure data not disclosed.” Do not invent a split.
  • Make the diligence ask specific: lease-start dates and original move-in dates for all units — not “more information on the tenancy.”
  • Never assume mark-to-market without a turnover schedule. A defensible stabilized number is in-place NOI plus vacant units at market plus contractually known rollover. Everything beyond that is an assumption someone should have to defend in writing.

The discipline costs you nothing except the ability to present a bigger upside number than the data supports. Buyers who price locked LTL as recoverable are paying the seller today for income that may arrive in a decade, or never.

This is the analysis our pipeline runs on every file.

Upload an OM, rent roll, or T-12 and get back a working Excel workbook with every number cited to its source page. First file free.

Try one free