Engineering-Based Cost Estimation
When actual costs are not available, studies estimate component costs from drawings, quantities, and published cost data. The IRS audit guide sets out what makes those estimates reliable. Published databases are estimating tools, not tax authority, and a uniform step-up to the purchase price is not an acceptable method.
Educational information, not tax advice.
Updated 2026-10-14
Key takeaways
- Estimates should come from contractors or reliable published sources, with the source cited.
- The same estimating technique should be used for all components.
- Estimates must reconcile to actual total cost; a uniform step-up factor is not acceptable.
- The audit guide names R.S. Means and Marshall Valuation Service as example sources, not as bindings.
- Residual estimation that prices only short-lived items can skew results toward those items.
The IRS audit guide's detailed engineering cost-estimate approach follows the same steps as a study based on actual cost records, but replaces actual costs with estimates. It names R.S. Means and Marshall Valuation Service as examples of reliable published sources, and expects the volume, page, and item to be cited.
In practice, an estimator takes quantities from drawings and the site inspection — linear feet of conduit, number of outlets, square feet of carpet — and applies unit costs adjusted for location and date. Indirect costs such as design fees, permits, and contractor overhead are then allocated.
The Service has not established standards for preparing studies. Substantiation using actual costs is more accurate than using estimates. Estimates exist because acquisition files rarely contain the original contractor’s unit prices.
CostSeg SI’s proposed estimation step is explanatory. This library does not claim that an RSMeans connection, extraction agent, or QA module is operational on live studies.
Educational information, not tax advice. An estimate that cannot be traced to a source and a quantity is not engineering support.
Engineering estimation is the usual method on acquisitions because the original contractor file is gone. It is a substitute for actual cost, not an improvement on it. When contemporaneous unit prices exist, the guide treats those prices as more accurate than a database line written years later.
Where estimates go wrong
The audit guide gives an example where an estimate built from a published database overstated the cost of electrical outlets compared with the contractor's actual unit cost. It also warns that mixing estimating techniques — detailed pricing for short-lived items and a residual for the building — can skew results toward shorter lives.
In Boddie-Noell Enterprises v. United States, the Court of Federal Claims found percentage estimates and after-the-fact extrapolations far too unreliable to support the claimed classifications.
The outlet example in the guide is a reminder that any estimate, from a published database or from a contractor, still needs applicable indirects. Matching a database line is not the end of the pricing step.
Reconciliation
Total allocated costs should reconcile to total actual costs or, for acquisitions, to the purchase price less land. The guide states that applying a pro-rata step-up to force reconciliation is not an acceptable method.
When the first pass of estimates misses the price, the difference needs an explanation: obsolescence, condition, or intangibles in the price. Scaling every line by the same factor hides that explanation and can inflate short-lived property.
Separately purchased furniture and equipment already on their own invoices should not be estimated again inside the building. Double-counting FF&E is a listed quality problem.
What published data is — and is not
Construction cost databases are estimating tools. The audit guide treats them as acceptable sources, not as authority, and the result is only as good as the quantities and adjustments applied to them.
The guide’s recognition of R.S. Means is the phrase “reliable published sources.” It does not endorse a vendor, lock a vintage of the database, or make a unit price a legal class.
Location factors and date adjustments matter. A national average unadjusted for the project city is a weaker estimate than the same line adjusted to the site and the placement year.
The six approaches, in order of detail
The guide lists six approaches and says others are derivatives: detailed engineering from actual cost records; detailed engineering cost estimates; contractor survey or letter; residual estimation; sampling or modeling; and rule of thumb.
Residual estimation prices only the short-lived assets and assigns the rest to the building. It is simpler and often less accurate, and it generally does not reconcile. Sampling is for near-identical facilities; the guide warns that a population under 50 can limit accuracy. Rule-of-thumb percentages with little documentation should be viewed with caution. Examiners are also told not to reject a rule-of-thumb study categorically, which is not the same as treating it as equivalent to a detailed take-off.
Quantity take-off
A take-off is a measurement: counts, areas, and lengths from drawings or from the inspection. Unit cost times quantity is the direct cost. Without a take-off, a database has nothing to multiply. Two estimators can use the same published source and still differ if one counted dedicated outlets and the other counted every receptacle in the building.
The guide lists determination of unit costs and engineering take-off as an element of a quality study. Photographs that support the quantities belong in the exhibits.
Sources
- 1IRS Publication 5653 — Cost Segregation Audit Techniques Guide (rev. 2-2025)(opens in a new tab)
Written for IRS examiners. The guide states it is not an official pronouncement of law and cannot be cited as authority. It predates the July 2025 law.
- 2