What a defensible 179D or 45L certification actually requires

Neither Section 179D nor Section 45L is self-certified, and neither is defended at the return level. When the IRS examines these claims — and its 179D practice unit tells examiners exactly what to request — the tax computation is rarely the issue. The certification file is. With both provisions past their June 30, 2026 termination triggers, the studies now being commissioned support in-flight projects and open-year lookbacks, which means they will be reviewed years from now by an examiner with the benefit of hindsight. They need to be built accordingly.

The 179D file

Section 179D(d) and Notice 2006-52 (as modified by later notices) set the floor:

• Qualified individual. An engineer or contractor licensed in the jurisdiction where the building sits — not merely licensed somewhere — and unrelated to the claimant under the Section 45(e)(4) standard. Independence matters most in designer allocation work, where study firms have been known to certify their own conclusions through affiliates.
• DOE-qualified software. The energy model must run on software from the Department of Energy's published qualified software list for the applicable year (the eQUEST, EnergyPlus, and IES-VE families dominate). A spreadsheet extrapolation is not a model.
• The right baseline. Savings are measured against the applicable ASHRAE 90.1 edition — 90.1-2007 for pre-2023 construction starts, 90.1-2019 for construction beginning 2023 through the sunset. Because 2019 is a far tighter reference, a study quoting 40 percent savings against the wrong edition overstates the sliding-scale rate, sometimes to zero.
• Field inspection. The certification must reflect an inspection of the building after it is placed in service, consistent with NREL guidelines, confirming the modeled systems were actually installed. Desktop-only studies are the single most common defect.
• Retention, not filing. The signed certification, the modeling inputs and outputs, square-footage support, cost support for the per-cost cap, prevailing wage and apprenticeship records where the enhanced rate is claimed, and the allocation letter for designer claims all stay in the file — complete before the return is filed, not assembled on audit.

The 45L file

45L substitutes program certification for modeling: the home must be certified under the applicable ENERGY STAR program or DOE's Zero Energy Ready Home program by an accredited rater (RESNET/HERS infrastructure, per Notice 2023-65), with the staged inspections those programs require — including at framing, which is why post-construction "certification" of a finished code-built home is generally impossible. The eligible contractor's file adds proof of ownership during construction, the acquisition date for each unit (dispositive after the June 30, 2026 acquisition cutoff — see the sunset brief), and prevailing wage records for enhanced multifamily amounts.

Red flags in cheap certifications

The market for these studies ranges from rigorous engineering practices to volume shops. The discount tier announces itself:

• No site visit, or a "virtual inspection" from photographs.
• Results promised before modeling — a proposal quoting the maximum per-square-foot rate from an address and a square footage.
• Wrong-state licensure or a certifier who is an employee or affiliate of the promoter.
• Baseline games: 90.1-2007 savings figures on 2023-or-later construction starts.
• Every building hits the cap. Genuine portfolios show a distribution across the 25–50 percent scale; uniform maxima suggest the model was tuned to the answer.
• Contingent fees on the claimed deduction — not per se improper, but correlated with all of the above, and a fact examiners notice.
• No workpapers delivered. If the taxpayer cannot obtain the modeling files, it cannot defend the claim when the promoter has moved on.

A cheap study does not merely risk the deduction; it risks accuracy-related penalties, and the certifier's independence problems are not curable retroactively. Given that a 179D deduction is an acceleration benefit in the first place — worth its time value, like cost segregation, not its face amount — paying for engineering that survives exam is not conservatism. It is the whole economics of the claim.