Cost segregation for short-term rentals: buckets and payback

By Max Medvedev · 6 Aug 2026

What a study does, in one paragraph

A cost-segregation study takes the price you paid and splits it into recovery classes: 5- and 7-year personal property, 15-year land improvements, and the long building shell. Everything with a MACRS life of 20 years or less is eligible for 100% bonus depreciation, so those short-life buckets can be expensed in year one. Land is in none of them.

The four buckets, and which ones bonus can reach

A typical study moves 20–30% of building basis into the 5- and 15-year classes. On a $1M property that is often $200,000–300,000 of first-year deduction once bonus applies.

The shell is 39 years, not 27.5, for most short-term rentals: a unit rented on a transient basis — average stay of 30 days or less — is not a dwelling unit under §168(e)(2), so the building is nonresidential. That fork and the bonus rules behind it are laid out in the depreciation guide, and the arithmetic that decides which side you land on is total nights ÷ reservations, never nights ÷ 365.

Bonus is 100% and permanent under OBBBA for property both acquired and placed in service after 19 January 2025, with acquisition meaning the written binding contract date. A contract signed 15 January 2025 caps the property at 40% no matter when the study is done.

Land is not a bucket, and land improvements are not land

Land is never depreciated. The purchase price splits between land and building before anything else, commonly by the county assessor's ratio or by the study's own allocation. Over-allocating to land throws away deduction; under-allocating invites challenge. A study does not shrink the land line — it divides what sits above it.

Land improvements are a separate class and are depreciable over 15 years. The driveway, the fence, the pool and the exterior lighting are what owners most often leave sitting in the 39-year bucket, and all of it is bonus-eligible.

What it costs, and the arithmetic that decides

An engineering-based study is quoted per property, and that quote is the one number this page cannot supply. Get it in writing first — everything below needs it as an input.

The payoff side is straightforward: first-year deduction × your marginal rate = cash saved this year. A physician earning $400,000 who buys a $1M short-term rental, runs a study, and materially participates is looking at $200,000–300,000 of first-year deduction — roughly $70,000–105,000 of cash tax saved in one year at a 35% marginal rate.

That is the top of the range, and four things sit between it and your bank account.

Four things that shrink the payoff

Material participation, or none of it lands this year. An average stay of seven days or less only removes the automatic passive label under Reg. §1.469-1T(e)(3)(ii)(A). It does not make the loss non-passive. Material participation is a separate requirement — usually 100 hours and more than any other individual, including your cleaner. Without it, a $250,000 deduction suspends as a passive loss and offsets nothing but future passive income. This is the largest single determinant of whether a study was worth commissioning, and it is decided by your hours log, not by the study.

The excess business loss cap. Even a fully non-passive loss is capped at $256,000 single / $512,000 married filing jointly for 2026 (Rev. Proc. 2025-32). The excess is not lost — it becomes an NOL carried forward indefinitely, usable against up to 80% of taxable income later, with no carryback. So a larger study does not automatically mean more cash in the current year, and there is a real conversation about electing out of bonus by asset class under §168(k)(7) to spread it.

The state may not follow. California does not conform to §168(k) bonus at all and caps §179 at $25,000. Federal and state depreciation then run as two separate schedules with add-backs, and the state bill claws back part of the federal win.

A December closing. Personal property normally uses the half-year convention, but if more than 40% of the year's personal property is placed in service in the fourth quarter, the mid-quarter convention applies to all of it. Real property is mid-month regardless. "Placed in service" means listed and genuinely available for rent, not the first guest's arrival.

§179 expensing is the other lever and the wrong one here: it cannot create a loss, which for a W-2 owner is the entire point. Bonus has no such limit.

Bought in an earlier year? Not too late.

An owner who bought without a study can still catch up. Form 3115, automatic consent DCN 7, filed with the current-year return, claims the cumulative missed acceleration as a one-time §481(a) deduction. No amended returns — the IRS requires the 3115 route rather than amendments.

The bonus rate stays fixed to the property's original placed-in-service year. A 2023 purchase catches up at 80%, a 2024 purchase at 60%, and a post-19-January-2025 purchase at 100%. The catch-up year does not upgrade the rate.

The exit is part of the price

The 5- and 7-year property a study creates is §1245 property, and its depreciation returns as ordinary income at sale, at rates up to 37%. The 15-year buckets and the shell are §1250, taxed as unrecaptured §1250 gain at a maximum of 25%. And a §1031 exchange no longer covers personal property, so the §1245 slice recaptures even inside an otherwise-deferred swap.

That makes a study partly a rate-and-timing trade rather than a pure discount — deductions now at your ordinary rate against recapture later. Depreciation recapture when you sell a short-term rental is the other half of the arithmetic, and it belongs in the decision before the study is ordered, not after.

Run it on your own numbers

Put your purchase price, closing date, and reservation mix through the estimator — it applies the 30-day building fork, the acquisition-date bonus rate, the loss cap, and the state add-back in order, with a citation behind every figure.