Companies increasingly lead with numbers the standards don't define: adjusted EBITDA, adjusted EPS, "core earnings," constant-currency growth. Some adjustments do inform you. A one-time plant fire really isn't part of run-rate economics. But the abuse potential is obvious. Define your own scoreboard and you win every quarter.
Are non-GAAP measures regulated?
The US rulebook is Regulation G plus Item 10(e) of Reg S-K. Any public non-GAAP measure must be reconciled to the most directly comparable GAAP measure, which must be presented with equal or greater prominence. Non-GAAP measures can't appear on the face of the GAAP statements, and labels can't mislead. Earnings releases land under 8-K Item 2.02 and follow the same rules. SEC comment letters police this actively. India has no codified non-GAAP regime. LODR prescribes the results format (Ind AS, with "exceptional items" on the face), but investor presentations are effectively unregulated, and "adjusted EBITDA" in an Indian deck answers to nobody. Several new-age listings popularized "adjusted EBITDA (excluding ESOP costs)". Treat the exclusion of stock compensation as unproven until you rebuild it. So your own habit has to work in both jurisdictions: never quote an adjusted number you haven't rebuilt yourself from the audited statements.
Advertisement
How do you classify non-GAAP add-backs?
The four-bucket add-back classifier takes every line in the reconciliation and assigns it:
- A: genuinely non-recurring, non-operating. Rare. Natural disaster losses, a one-time legal settlement for a discontinued product. Excluding them is legitimate if both gains and losses go, so check that they exclude gains too.
- B: "one-timers" that recur. Restructuring charges appearing 4 years out of 5, where serial restructurers are simply booking operating costs in a costume. "Integration costs" at a company whose strategy is acquiring. Annual "exceptional" inventory write-offs.
- C: real economic costs defined away. SBC is the flagship. It's a real transfer of value from you to employees and ~always added back, as the stock-based compensation page showed. Amortization of acquired intangibles at a company whose growth is bought belongs here too, since excluding it pretends acquisitions were free. So do "pre-opening costs" at a chain that always opens stores.
- D: presentation variants. Constant-currency growth, organic growth. These are fine if reported symmetrically in good and bad years, so check whether constant-currency disappears when FX turns favorable.
The wedge test and two real cases
Compute the wedge test every time. Cumulative 5-year adjusted net income − cumulative 5-year GAAP net income = the wedge. Express it as % of cumulative GAAP earnings and % of market cap. A wedge of 10–15% is life, but 40%+ means the adjusted series is a different company. Complete the picture with SBC as % of revenue and the diluted share count trend. If "adjusted profit" grows while share count grows 3%+ annually, per-share owners may be standing still.
Two named specimens are worth carrying. The figures are approximate, as publicly reported, so verify them against the original documents. WeWork's 2018 bond prospectus introduced "community-adjusted EBITDA." It turned 2017 revenue of ≈$886m and a net loss of ≈$933m into a positive ≈$233m by excluding even building-level marketing and admin, and it became the genre's warning label. Valeant pushed "cash EPS" near $10 while GAAP results collapsed toward losses (2015 onward). Most of the gap was amortization of the very acquisitions that made up the strategy, which puts it in bucket C. Schilit's taxonomy, the spine of financial shenanigans, treats non-GAAP gaming as its own shenanigan family. For now, build the habit: find the bridge, rebuild the bridge, classify every plank.