Cost Segregation for Large Multifamily (100+ Units): Institutional-Grade Studies

At 100 units and above, cost segregation stops being a one-time deduction exercise and becomes part of ongoing fixed asset management. The study still reclassifies components under IRC Sec. 168(e), but the methodology, the documentation burden, and the follow-on planning all change.

Reclassification percentages typically compress slightly at this scale, landing between 19% and 25% of depreciable basis. Larger buildings carry more structure per dollar and less site improvement per unit, especially when parking moves from surface lots into a structured garage.

Statistical Sampling Replaces Full Enumeration

Enumerating every outlet, fixture, and cabinet across 240 units is not economically rational, and the IRS does not require it. Rev. Proc. 2011-42 provides the framework for statistical sampling in fixed asset studies. A properly designed sample of unit types, with documented stratification and confidence intervals, supports extrapolation across the property.

This is the single largest methodological difference between a small property study and an institutional one. It is also where cheap studies fail. A sample that is not stratified by unit type, floor plan, and renovation vintage does not support the extrapolation, and an examiner who identifies that flaw can challenge the entire reclassification.

Structured Parking Changes the Math

Surface parking is a 15-year land improvement. A structured parking garage that is part of the building is generally structural and depreciates with the building over 27.5 years. A freestanding garage may be analyzed separately.

This is why a 200-unit urban mid-rise with a podium garage often reclassifies 19% while a 200-unit garden complex with surface parking reclassifies 25%. Same unit count, materially different result. Investors comparing acquisition targets should not assume a uniform depreciation benefit across property formats.

Worked Example: 180 Units

An institutional buyer acquires a 180-unit property for $31,000,000. Land is allocated at $4,650,000, leaving $26,350,000 depreciable. The study identifies five-year property of $2,898,500 (11%), seven-year property of $395,250 (1.5%), fifteen-year land improvements of $2,371,500 (9%), and structure of $20,684,750 (78.5%).

Reclassified basis of $5,665,250 is deductible in year one under IRC Sec. 168(k). Structural depreciation adds roughly $752,175. Total first-year depreciation is approximately $6,417,425, against $958,182 straight-line.

The Partial Disposition Election Is Worth More Than the Study

Large properties get renovated. When you replace 180 roof sections, re-pipe a building, or gut and rebuild unit interiors during a value-add program, the old components are still sitting on your depreciation schedule.

The partial disposition election under Treasury Regulation Sec. 1.168(i)-8(d) lets you write off the remaining basis of the retired component in the year of disposal. You cannot make that election without knowing the basis of the retired component, and you cannot know that basis without a cost segregation study that broke the building into components in the first place.

For a value-add sponsor planning $4 million of interior renovation over three years, the partial disposition elections enabled by the initial study frequently exceed the value of the study's own acceleration. This is the argument for running the study at acquisition even when the sponsor cannot immediately use the loss.

Passive Loss Reality for Institutional Investors

The larger the deduction, the more likely it exceeds available passive income. Under IRC Sec. 469, suspended losses carry forward indefinitely and free up on a fully taxable disposition of the activity.

For a fund holding multiple properties, grouping elections under Treasury Regulation Sec. 1.469-4 become important. Properly grouped activities let income from stabilized assets absorb losses from newly acquired ones. That election has consequences on disposition, since a grouped activity is not fully disposed of until every property in the group is sold, so it should be made deliberately rather than by default.

Investors evaluating whether the deduction is usable at all should start with our overview of using cost segregation against W-2 income and the broader complete guide to cost segregation.

Frequently Asked Questions

Does the IRS accept statistical sampling on large properties?

Yes. Rev. Proc. 2011-42 provides the framework for sampling in fixed asset determinations. The sample must be properly designed, stratified by unit type and vintage, and documented with confidence intervals. Sampling without that documentation is the vulnerability, not sampling itself.

Why does my urban mid-rise reclassify less than a garden complex?

Structured parking. A podium or subterranean garage integrated with the building is generally structural and depreciates over 27.5 years, while surface parking is a 15-year land improvement. That single difference commonly moves the total reclassification percentage by five or six points.

What is a partial disposition election and why does it need a cost seg study?

It lets you deduct the remaining basis of a building component you remove, under Treas. Reg. Sec. 1.168(i)-8(d). You can only do that if you know the component basis, which requires a study that broke the building into components. For value-add sponsors this is often worth more than the original acceleration.

Can a fund group properties to use the losses?

Grouping elections under Treas. Reg. Sec. 1.469-4 can let stabilized property income absorb new acquisition losses. The tradeoff is that a grouped activity is not fully disposed of until all properties in the group are sold, which delays release of suspended losses on individual sales.

How long does a 180-unit study take?

Typically three to five weeks. The site work, sampling design, and cost detail reconciliation take longer than a small property, and institutional buyers usually want the workpapers assembled to a standard that supports examination years later.

---