To calculate normalized earnings per share, start with reported GAAP net income, add back each non-recurring charge on an after-tax basis, subtract each non-recurring gain on an after-tax basis, and divide the result by diluted shares outstanding. The point is to isolate what the business earns on a repeatable basis, stripping out one-time items like restructuring charges, asset sale gains, litigation settlements, and impairments that would otherwise distort any P/E ratio, ROE calculation, or valuation model you build on top of the reported number.
The Formula and the Four Steps
The calculation itself is short:
Normalized EPS = (GAAP net income ± after-tax non-recurring items) ÷ diluted shares outstanding
Four steps get you there.
- Pull GAAP net income and diluted share count from the income statement.
- Identify each non-recurring item hitting that period’s results.
- Tax-effect each item using the company’s effective tax rate.
- Add back non-recurring charges, subtract non-recurring gains, then divide by diluted shares.
Most analysts begin from diluted EPS rather than basic, because the diluted figure reflects the full capital structure including options, RSUs, convertibles, and warrants the company has already promised. The diluted share count is always equal to or greater than the basic count.
Both diluted and basic EPS share the same weakness that normalization exists to fix: they rely on GAAP net income, which captures a factory sale, a legal settlement, an asset write-down, and normal product revenue in a single line.
Which Items to Strip Out
The entire exercise hinges on sorting items into recurring and non-recurring buckets. Get this wrong and you build a flattering version of earnings rather than an accurate one. The following items are the usual candidates, and companies disclose them in the Management Discussion and Analysis section and the footnotes of the 10-K or 10-Q.
- Restructuring charges. Severance, facility closures, and contract termination costs tied to a specific reorganization.
- Asset sale gains or losses. Selling a factory, a business unit, or a large investment position affects net income without touching the operating engine.
- Litigation settlements and regulatory fines. A class-action payout crushes reported earnings; the release of a legal reserve no longer needed produces a one-time gain that should also come out.
- Goodwill and intangible asset impairments. Large non-cash charges that reduce reported income without affecting cash flow.
- Discontinued operations. GAAP reports these separately, but the results still flow into the net income line used for EPS.
- Accelerated depreciation and early write-offs. When equipment is retired ahead of schedule, remaining book value expenses at once.
Management often flags what it considers non-recurring. Take that framing with skepticism. If a company runs a “restructuring” program in three or more of the last five years, those charges are a recurring cost of doing business and belong in earnings, not in the adjustment column.
Tax-Effecting Each Adjustment
This is the step first-time analysts miss. Non-recurring items appear on the income statement before taxes, meaning they already moved the company’s tax bill. You cannot add back a $10 million restructuring charge dollar-for-dollar, because the charge saved the company money on taxes. Only the after-tax impact gets added back.
The formula for any single adjustment:
After-tax adjustment = Pre-tax item × (1 − effective tax rate)
A $10 million restructuring charge at a 25% effective tax rate reduced net income by $7.5 million. The other $2.5 million showed up as lower tax expense. So the add-back is $7.5 million, not $10 million.
The same logic works in reverse for gains. A $5 million pre-tax gain on a building sale increased net income by $3.75 million after tax. Subtract $3.75 million.
Use the company’s effective tax rate, not the statutory federal rate. The effective rate, disclosed in the income tax footnote of the 10-K, accounts for state taxes, foreign obligations, credits, and permanent differences between book and taxable income. When a specific non-recurring item has its own tax treatment disclosed in the footnotes (a domestic asset sale taxed at capital gains rates, for example), use the item-specific rate.
One important exception: goodwill impairments are often non-deductible for tax purposes, meaning the company received no tax benefit from the charge. In that case, add back the full pre-tax amount. Check the footnotes to confirm.
A Worked Example
Take a company reporting the following for the year:
- GAAP net income: $400 million
- Diluted shares outstanding: 200 million
- Reported diluted EPS: $2.00
- Pre-tax restructuring charge: $50 million
- Pre-tax gain on sale of a business unit: $20 million
- Goodwill impairment (non-deductible): $30 million
- Effective tax rate: 24%
Tax-effect each item:
- Restructuring charge add-back: $50M × (1 − 0.24) = $38.0 million added back
- Asset sale gain removal: $20M × (1 − 0.24) = $15.2 million subtracted
- Goodwill impairment add-back: full $30 million, because there was no tax benefit
Net adjustment: +$38.0M − $15.2M + $30.0M = +$52.8 million
Normalized net income: $400M + $52.8M = $452.8 million
Normalized EPS: $452.8M ÷ 200M shares = $2.26
Reported EPS of $2.00 understated recurring earning power by about 13%. Anyone using the reported figure to calculate a P/E ratio would overestimate how expensive the stock is relative to its sustainable profits.
When One Year Isn’t Enough: Cyclical Smoothing
Removing one-time items is half the picture. For companies in cyclical industries like steel, homebuilding, autos, or energy, even “clean” earnings swing wildly with the economic cycle. A homebuilder’s earnings at the peak of a housing boom are not representative of what it earns on average, no matter how carefully you strip out charges.
The standard fix is averaging across a full cycle, typically five to ten years spanning at least one peak-to-trough-to-peak sequence. Two methods dominate:
- Average dollar earnings. Sum adjusted net income across each year of the cycle and divide by the number of years. Works best when the company hasn’t changed materially in size.
- Average margin applied to current revenue. Calculate the average operating margin across the cycle and apply it to today’s revenue base. This handles growth better, because a company that doubled in size over seven years would have its mid-cycle earnings distorted by a simple dollar average.
The Shiller CAPE ratio applies this concept at the market level, averaging ten years of inflation-adjusted earnings. The same principle applies to individual companies. When an industrial name trades at a seemingly low P/E, check whether current earnings sit near a cyclical peak. Mid-cycle normalized EPS often tells a different valuation story than a single year’s result.
The Stock-Based Compensation Judgment Call
Stock-based compensation is the most contentious normalization item in practice. Companies routinely exclude it from their adjusted earnings. In the S&P 1500 technology sector, roughly two-thirds of companies strip it out. The argument is that the expense is non-cash and that any dilution already shows up in the diluted share count.
That argument has a hole. Stock compensation transfers ownership from existing shareholders to employees. Whether the cost arrives as cash salary or as equity that dilutes your stake, it is real compensation for real work. Ignoring the expense while counting the dilution gives a partial picture. Many independent analysts treat stock-based compensation as a genuine operating expense and leave it in normalized earnings, particularly for technology companies where it can run 15% to 25% of operating income.
There is no universally correct answer. Be consistent. If you strip stock-based compensation from one company, do the same for every peer in the comparison set.
Checks Against Fooling Yourself
The biggest risk in normalization is turning it into a flattery machine. Every charge you remove makes earnings look better, and there is always a plausible argument for excluding one more item. A few hard rules keep the exercise honest.
Apply adjustments symmetrically. If you add back a litigation loss, you must also remove any litigation gains. The SEC specifically calls out one-sided adjustments as a red flag for misleading non-GAAP measures.1SEC.gov. Non-GAAP Financial Measures Excluding a charge in one period but not similar charges in prior periods raises the same problem.
Watch for chronic “one-time” items. Restructuring charges that appear in three or more of the last five years belong in earnings, not in the adjustment column.
Document each adjustment and its rationale. This prevents drift. An exclusion made for a clear one-time event three years ago can quietly become a standing exclusion with no defensible basis.
Compare your figure against the company’s own adjusted EPS. Significant gaps in either direction deserve investigation. If your number is materially higher than management’s, you may be too aggressive. If it is lower, the company may be excluding items that are more recurring than its reconciliation table admits. Public companies that disclose a non-GAAP measure must present the most directly comparable GAAP measure and a quantitative reconciliation between the two, under Regulation G.2eCFR. 17 CFR Part 244 – Regulation G Read that reconciliation line by line.
Using Normalized EPS in Valuation
The most common application is a more stable price-to-earnings ratio. In the worked example above, a $70 stock price against reported EPS of $2.00 produces a P/E of 35x. Against normalized EPS of $2.26, the P/E is roughly 31x. That gap can change an investment decision, especially when compared against the company’s historical average or its sector peers.
Peer comparison is where normalization earns its keep. Two companies in the same industry might report wildly different EPS in a given year because one sold a division and the other absorbed a legal settlement. Comparing their reported P/E ratios tells you almost nothing about relative value. Normalizing both strips the noise and puts operating performance on level ground.
Normalized net income also feeds directly into return on equity. A one-time gain inflates the numerator and makes management look more efficient than it is; a one-time charge does the opposite. Using the normalized figure gives a cleaner read on how well the company deploys shareholder capital under normal conditions.
For longer-range work, normalized EPS provides a stable base for growth projections in a discounted cash flow model. Projecting five years of growth from a single year that included a massive restructuring charge will understate the trajectory of the underlying business. Starting from a normalized figure anchors the model to what the company can actually generate across a full operating cycle.