Cost Segregation

Cost segregation is the analysis that breaks a building's total cost into the separate items of property inside it, so that each item is depreciated over its own recovery period rather than all of it being written off over the 27.5 or 39 years the structure itself requires. A study separates land, which is never depreciated, from land improvements, from the tangible personal property that sits in and around the building, from the structural shell.

The IRS does not treat "cost segregation" as a legal term, and it names the exercise slightly differently. Its Cost Segregation Audit Technique Guide (Publication 5653, rev. 2-6-2025) explains that where only lump-sum costs are available, estimating techniques may be needed to "segregate" or "allocate" costs to individual assets, and that "this type of analysis is generally called a 'cost segregation study,' 'cost segregation analysis,' or 'cost allocation study.'" So the study is the deliverable and cost segregation is what the study does. Nothing in the Internal Revenue Code uses either phrase; what the Code supplies is the classification rules the study applies.

Why the split is worth money. Publication 5653 states the mechanism compactly. A building, "termed '§ 1250 property,'" is generally nonresidential real property with a 39-year recovery period or residential rental property with a 27.5-year recovery period, and it "must use straight-line depreciation." Equipment, furniture and fixtures, "termed '§ 1245 property,'" are tangible personal property, which has a shorter recovery period and can also be eligible for accelerated depreciation. IRS Publication 946 supplies the classes: 5-year and 7-year for tangible personal property, and 15-year for "certain improvements made directly to land or added to it (such as shrubbery, fences, roads, sidewalks, and bridges)" and for qualified improvement property placed in service after 2017. Land itself has no recovery period at all, which is why a study that moves cost out of the land column is doing the opposite of a favor.

Where the classification rules come from, which explains why the answers are argued about. Publication 5653 records that allocations "are typically based on criteria established under the Investment Tax Credit (ITC) laws under § 48," a credit repealed decades ago, and that "complex and often conflicting guidance relating to property qualifying for ITC, resulting from numerous legislative acts, court decisions and Service rulings, and a lack of bright-line tests, have impacted the ease of distinguishing § 1245 property from § 1250 property." The guide is blunter still about the practical consequence: "the allocation of building components to § 1245 property is often a contentious issue." An owner reading a study that presents its percentages as settled facts is reading a position, not an answer.

What the audit guide is, and what it is not. Publication 5653 says of itself that it "is not an official pronouncement of the law or the position of the Service and cannot be used, cited, or relied upon as such." That makes it a poor source for a legal conclusion and an excellent source for something else: it is the IRS's own account of what an examiner looks for. It sets out six approaches examiners expect to recognize, from a detailed engineering approach built from actual cost records down to a "rule of thumb" approach, and it lists 13 principal elements of a quality study, including preparation by an individual with expertise and experience, a detailed description of the methodology, an explanation of the legal analysis, reconciliation of total allocated costs to total actual costs, and consideration of related aspects including the change in accounting method. It also states plainly that "there are no prescribed qualifications for cost segregation preparers." Nothing stops anyone from selling a study, and the report either shows its work or it does not.

The look-back study is the part almost every summary gets wrong. A study run on a building that was placed in service in an earlier year, and already depreciated as a whole, does not produce amended returns. Publication 5653's own explanation cites live authority for this: once a method of accounting has been adopted, "the taxpayer may not change the method by amending its prior income tax returns" (Rev. Rul. 90-38, 1990-1 C.B. 57), so amended returns or refund claims based on a later study "should generally be disallowed on the basis that the taxpayer is attempting to make a retroactive method change." The governing provision is section 446(e), which requires a taxpayer changing a method of accounting to "secure the consent of the Secretary" before computing income under the new one.

The regulation is unusually helpful here, because its drafters chose this exact fact pattern for one of their own examples. 26 CFR 1.446-1(e)(2)(iii), Example 9, describes a taxpayer that placed a building in service in 2003, classified the whole $10,000,000 as nonresidential real property, and in 2006 completed "a cost segregation study on the building and its components" that identified $1,500,000 as section 1245 property. The regulation's conclusion: "A1's change to this depreciation method, recovery period, and convention is a change in method of accounting. This method change results in a section 481 adjustment." Publication 538 lists "a change in the depreciation or amortization method" among the changes that require IRS approval, and the application is made on Form 3115.

The practical effect of a section 481(a) adjustment is that the entire catch-up, the difference between the depreciation actually claimed in the earlier years and the depreciation the corrected classification would have produced, is taken into account in the year of the change rather than spread back across the returns already filed. That single-year concentration is the reason look-back studies are marketed at all, and it is also why the timing of a study interacts with everything else on the return in that year.

The sequence for an owner is short, and the order matters.

1. Establish the depreciable basis and strip out land. Land is never depreciated, so the study's starting point is the building and improvements, not the purchase price.

2. Identify the components and price them. This is the engineering half: the study assigns costs to individual assets or asset groups sharing a recovery period and a placed-in-service date, either from actual cost records or by estimating from construction data.

3. Classify each component. Section 1245 tangible personal property, land improvements, or part of the section 1250 structure. This is the step that is argued about.

4. Reconcile. A quality report ties total allocated costs back to total actual costs, so nothing has been created or lost in the allocation.

5. Pick the mechanism. For a building placed in service in the current year the new classification simply goes on the return. For an earlier year it is a Form 3115 change of method with a section 481(a) adjustment.

A hypothetical example of the catch-up, simplified so the arithmetic is visible: assume the building went into service on January 1 and ignore the part-month convention that would trim the first year slightly.

Marcus bought a small commercial building three full years ago and treated the entire depreciable cost as nonresidential real property. A study now identifies $500,000 of that cost as 5-year property.

What was claimed on that slice. Over 39 years, straight line, $500,000 produces $500,000 ÷ 39 = about $12,821 a year, so $500,000 × 3 ÷ 39 = $38,462 across the three years.

What should have been claimed. Under the general depreciation system, 5-year property using the 200 percent declining balance method and the half-year convention runs at the published rates of 20.00 percent, 32.00 percent and 19.20 percent in years one to three. On $500,000 that is $100,000 + $160,000 + $96,000 = $356,000.

The adjustment. $356,000 − $38,462 = $317,538. That is the section 481(a) adjustment, and it is taken into account in the year of the change on Form 3115, not by reopening the three earlier returns.

Two things follow that the arithmetic does not show. The $317,538 is a deduction the owner may not be able to use in full this year, because a rental loss runs into the passive activity loss rules before it reaches other income. And the $500,000 is still basis, so the acceleration is borrowed against a larger recapture on sale.

Pros

• It converts a slow deduction into a fast one without changing the total, which is worth real money whenever the owner's time value of money is positive.
• The section 1245 slice can reach accelerated methods, bonus depreciation and section 179 expensing, none of which the 27.5-year or 39-year structure can use.
• A look-back study concentrates several years of missed depreciation into a single section 481(a) adjustment in the year of the change, rather than forfeiting it.
• It also produces an asset schedule, which is what makes a later partial disposition or a repair-versus-improvement question answerable.

Cons

• The classification criteria trace to a repealed credit and the IRS's own guide calls the allocation of building components to section 1245 property "often a contentious issue," so the answer a study gives is a position that can be examined.
• There are no prescribed qualifications for preparers, so the quality of a study is not guaranteed by anything except its own documentation.
• Nothing is forgiven. Every dollar accelerated reduces basis and comes back on disposition, and the section 1245 portion comes back as ordinary income.
• The resulting larger rental loss may be suspended rather than deducted, which can defer the benefit for years.
• A study costs money and takes professional time, so on a small depreciable basis the fee can consume most of the benefit.
• Getting the mechanism wrong is expensive: amended returns based on a later study should generally be disallowed, so a study filed the wrong way buys nothing.