Why R&D Expensing Makes Earnings Look Smaller: What Founders Need to Know

R&D expensing makes reported profits understate earning power, Dino Palazzo of the Federal Reserve Board argues in a 2026 working paper; with special items, it helped cut GAAP earnings from 82.8% to 66.4% of adjusted earnings. What FASB Statement No. 2 requires, what the data show, and how to explain an R&D-heavy loss.

R&D expensing makes reported profits understate the earning power behind them, and the gap has widened as research spending grew, according to research by Dino Palazzo, a principal economist at the Federal Reserve Board, in a paper dated September 18, 2026. In its market-wide data, R&D rose from 16.0% of adjusted earnings before December 1991 to 21.9% afterward, and GAAP earnings, squeezed by expensing and by a jump in special items, fell from 82.8% of that adjusted figure to 66.4%.

Where it stands: The paper’s sample ends in 2025; its later period runs from December 1991 through that year. As of late 2025, it finds stock valuations elevated on both the traditional and the corrected measure. It says nothing about 2026 results or about private companies.

The paper is about the stock market’s CAPE ratio, the cyclically adjusted price-earnings measure Robert Shiller developed with John Campbell. Its accounting argument, though, is the part founders meet every time they put a P&L in front of an investor. Below: what the rule does, what the paper measured, and how to explain a research-heavy loss without overselling an adjusted number. This is general information, not accounting advice.

What R&D expensing means

The rule comes from FASB Statement No. 2, “Accounting for Research and Development Costs,” dated October 1974 and in force for fiscal years starting on or after January 1, 1975. Its core requirement fits in one sentence: “All research and development costs encompassed by this Statement shall be charged to expense when incurred.” Palazzo’s paper sums the standard up as having “mandated immediate expensing of R&D rather than capitalization.”

The same statement draws the line against physical assets. Equipment and facilities acquired for research that have alternative future uses are recorded as tangible assets, and it is their depreciation that counts as R&D cost each year. Salaries of research staff are R&D costs, expensed as incurred, and so are materials and equipment bought for a single project with no other use. In practice, a company that builds a plant spreads the cost over the plant’s working life, while a company that pays engineers the same amount to build a new product takes the whole cost in the year it spends it.

FASB’s reasoning, laid out in the statement’s basis for conclusions, rests on uncertainty. It cites a study finding that “less than 2 percent of new product ideas and less than 15 percent of product development projects were commercially successful,” and concludes that capitalizing research costs “is not useful in assessing the earnings potential of the enterprise.” The Board weighed four approaches, including capitalizing every cost and capitalizing only costs that met specified conditions, before choosing to expense it all.

Palazzo’s paper comes at the same choice decades later and from a different angle. Its concern is comparability over time: as more of the economy’s earning power went into research, immediate expensing turned reported profit into a shrinking share of what the paper treats as underlying profitability.

How far reported earnings fell behind

The paper defines adjusted earnings as “GAAP earnings before R&D expenditures and special items” and tracks them against reported earnings from 1957 to 2025. The gap opened in steps: “the proportional gap averages 9% before 1974, increases to about 20% over 1974–1984, and reaches approximately 31% by 1991.”

Measure (paper’s Table 2) 1957 to Nov. 1991 Dec. 1991 to 2025
R&D, % of adjusted earnings 16.0% 21.9%
Special items, % of adjusted earnings 1.2% 11.7%
GAAP earnings, % of adjusted earnings 82.8% 66.4%
Traditional CAPE 15.2 27.6
CAPE-H (corrected) 14.7 19.3

The paper links the R&D increase to “the economy’s shift toward technology, pharmaceuticals, and other R&D-intensive firms.” The rise in special items has a separate cause. Accounting changes in the mid-1990s, among them EITF Issue 94-3 on restructuring exit costs and SFAS No. 121 on impairments of long-lived assets, standardized how two common kinds of special item were booked, and recognition of those charges climbed in the years after. Measured as shares of adjusted earnings, the table’s change column shows special items up 10.5 points and R&D up 5.9, so the fall in GAAP earnings is not R&D’s doing alone.

The valuation effect runs through the denominator. When R&D is “roughly 20% of adjusted earnings,” the paper argues, “reported earnings persistently understate sustainable profitability, mechanically elevating the measured accounting P/E ratio holding prices and underlying economics fixed.” Together with special items, that is why traditional CAPE in the table sits so far above the corrected version after 1991.

Why GAAP earnings got noisier too

Expensing lowers the level of reported profit. Special items make it jumpy. The paper compares earnings volatility with sales volatility, and for GAAP earnings that ratio “rises from 1.77 in 1957–1979 to 3.47 in 1980–2002 and 7.02 in 2003–2025, a nearly fourfold increase.” For adjusted earnings it moves only from 1.53 to 1.60 and then to 2.55.

The noise shows up in payout ratios, which set what companies return to shareholders against what they report as profit. The paper’s mean GAAP payout ratio “rises from 0.54 in 1957–1979 to 0.80 in 1980–2002 and 1.09 in 2003–2025.” Its starkest case is 2008, when “aggregate GAAP earnings fell by 78%, while total payout declined by only 32%, mechanically pushing the payout ratio to 4.30.” Payouts fell far less than the accounting profit they are measured against.

For a founder, the point is what one year of GAAP profit can and cannot tell an outsider. A write-down or restructuring charge can swing net income in ways that say little about the years ahead, and a year of heavy research can make a growing company look worse than a rival that has stopped investing.

Why investors look at earnings before R&D and special items

The paper adds R&D back and takes special items out for consistency, not generosity. In the ten-year earnings averages behind CAPE, it says, R&D “creates the dominant level shift,” while special items add a smaller shift plus swings concentrated in downturns. When it tests which correction does more for its out-of-sample return forecasts, it credits about 73% of the gain to R&D and 27% to special items.

Two cautions keep this from turning into permission to ignore costs. First, the paper calls its own adjustment “a measurement-consistency correction rather than an attempt to model economic earnings.” It is not claiming R&D is free or that adjusted profit is cash. Second, FASB’s 1974 argument is a fair counterweight: the payoff from research is uncertain project by project. An investor who adds R&D back is asking a narrower question, roughly what the business earns before its bets on future products, and still has to judge whether those bets will pay.

Explaining an R&D-heavy loss to investors

None of this changes how your books are kept; that turns on the standards that apply to you and on your auditor. It does say something about presentation. Some general practices that follow from the paper and the rules above:

  • Give R&D its own line. Statement No. 2 required disclosure of “the total research and development costs charged to expense in each period for which an income statement is presented.” A separate line lets an investor see how much of a loss is spending on what comes next.
  • Be strict about what counts. The statement’s list of activities that are not R&D includes “Routine, on-going efforts to refine, enrich, or otherwise improve upon the qualities of an existing product.” Labelling maintenance engineering as research inflates any before-R&D figure, and it is the kind of thing a due diligence review can unpick.
  • Put the GAAP number next to any adjusted one. For SEC registrants, Regulation G requires any published non-GAAP measure to come with “the most directly comparable financial measure calculated and presented in accordance with Generally Accepted Accounting Principles (GAAP)” and a reconciliation between the two. A private company’s investor update is a different setting, but the same pairing is a sensible standard for a board deck.
  • Keep cash apart from the label. Expensed or capitalized, research dollars leave the bank account at the same time, so runway planning runs on cash, not on adjusted earnings. Our guide to startup runway before Q4 works through that side.

The same discipline applies to the top line. Our explainers on gross vs net revenue in partner sales and annualized revenue vs ARR show how a presentation choice can move a headline number without changing the business; R&D expensing is that problem on the cost side. Investors will look at both views, so it helps to know which metrics venture investors weigh before picking the earnings figure to lead with.

What the paper does not tell founders

The paper is a study of public-market valuation. The figures above are market-wide aggregates, not results for individual firms, and the paper is not about startups, private companies or venture rounds. It is not accounting guidance for any company, and its adjusted earnings figure is a research construct, not a GAAP measure. Tax treatment of research spending follows separate rules and is outside this article.

It is also one economist’s working paper. Its author is on the research staff of the Federal Reserve Board, and the paper states that its views “do not necessarily reflect the views of the Board of Governors of the Federal Reserve System or its staff.” Its market finding, that both the traditional and the corrected CAPE stood above the 97th percentile of their history in late 2025, describes the stock market at that date, not any company’s prospects.

Frequently Asked Questions

Is R&D expensed or capitalized?

Under FASB Statement No. 2, dated October 1974, research and development costs within its scope are expensed when incurred. Equipment and facilities bought for research that have alternative future uses are capitalized instead, and their depreciation counts as R&D cost. Whether a given cost is in scope depends on the statement’s definitions, so check specific items with your accountant.

Why does R&D expensing make earnings look smaller?

The full cost of research lands on the income statement in the year it is spent, while spending on physical assets is spread over their useful lives through depreciation. Research by Dino Palazzo of the Federal Reserve Board finds R&D rose from 16.0% of adjusted earnings before December 1991 to 21.9% afterward, and argues that at around 20% reported earnings persistently understate sustainable profitability.

What are special items in earnings?

They are charges such as write-downs, asset impairments and restructuring costs. The paper finds special items rose from 1.2% of adjusted earnings before December 1991 to 11.7% afterward, following mid-1990s accounting changes, and that they add volatility concentrated in downturns.

Should a startup show investors earnings before R&D?

That is a call for you, your board and your accountant. If you do show such a figure, the common discipline is to label it clearly, show the GAAP number beside it and reconcile the two; SEC Regulation G requires the last two of registrants that publish non-GAAP measures. Investors will still weigh cash burn and whether the research will pay off.

Why did FASB choose expensing over capitalizing R&D?

In Statement No. 2, the Board weighed four methods, including capitalizing all costs and capitalizing only costs that met specified conditions. It chose expensing because the future benefits of individual research projects are highly uncertain and hard to measure, and it concluded that capitalizing them would not help users assess a company’s earnings potential.

Sources

This article is general information for founders and is not accounting, tax, legal or investment advice. How your company treats R&D depends on your facts and the standards that apply to you; work through it with your accountant or auditor. It summarizes a working paper by a Federal Reserve Board economist, whose views are the author’s own and not the Board’s.