Cohan v. Commissioner: what the estimation doctrine does and does not save

Cohan v. Commissioner, 39 F.2d 540 (2d Cir. 1930), is the most generous rule in tax procedure, and the most oversold. George M. Cohan — the Broadway producer and songwriter — spent freely on travel and entertainment but kept almost no records; the Board of Tax Appeals disallowed the deductions entirely. Judge Learned Hand's Second Circuit reversed in part: where the court is convinced that deductible spending genuinely occurred, absolute disallowance is wrong, and the court should make as close an approximation as it can — "bearing heavily if it chooses upon the taxpayer whose inexactitude is of his own making." That sentence is both the doctrine and its warning label.

The dispute and the holding

Cohan deducted tens of thousands of dollars of travel and entertainment expenses incurred producing and promoting his shows. The spending was plainly real — witnesses and the nature of his business established that — but he could not produce receipts or itemized records, and the Board disallowed the lot. The Second Circuit held that total disallowance was error: once the trier of fact is persuaded that some deductible amount was spent, it must estimate that amount rather than treat it as zero. But the estimate need not be charitable. The court may resolve every uncertainty against the taxpayer, since the evidentiary gap is the taxpayer's own doing.

The limits that matter

The doctrine has three hard edges. First, it estimates amounts, not entitlement: the taxpayer must still prove that a deductible event occurred and supply some rational basis — testimony, patterns, partial records — from which a figure can be derived. Courts routinely refuse to guess where the record offers no anchor at all. Second, Congress overrode Cohan where it mattered most to Cohan himself: Section 274(d) of the Internal Revenue Code imposes strict substantiation for travel, meals, gifts, and listed property — no adequate records, no deduction, no estimate. Third, even where Cohan survives, the bearing-heavily principle means the estimated number is usually a fraction of the claim.

What it means for specialty claims today

Cohan is regularly invoked, and regularly fails, in the specialty-tax setting. A cost segregation position is not a single spending total; it is hundreds of asset-level classifications, each turning on engineering facts — how an asset is affixed, what it serves, what it cost. A court cannot estimate its way to the conclusion that particular wiring serves particular equipment, which is why thin studies get reclassified wholesale, as in AmeriSouth XXXII v. Commissioner, T.C. Memo 2012-67, rather than salvaged by approximation. The IRS's Cost Segregation Audit Techniques Guide presumes contemporaneous, methodical support, and depreciation classes must still be assigned under the rules of Pub 946 whatever the estimate. Research credit claims fare no better: courts have declined to estimate qualified research expenses where the taxpayer cannot tie wages or supplies to qualified activities with credible evidence.

The honest use of Cohan is as a backstop for gaps at the edges of a well-documented claim — a missing invoice within a proven project, not a missing study. Planning on it is planning to lose most of the number. The reliable alternative is unglamorous: build the record before the return is filed. See audit readiness and documentation for what that record looks like.

Related reading

• How cost segregation became law — the doctrines that require real documentation
• AmeriSouth XXXII v. Commissioner — what happens when the proof never arrives
• What is cost segregation? — the documented alternative to estimation