Stock valuations · Analysis
R&D expensing and write-downs: how they distort P/E ratios
GAAP counts research as a cost and books large one-time charges, so reported earnings can understate profits. See how that pushes P/E and CAPE ratios higher.
A P/E ratio is only as good as its E
A price-to-earnings ratio divides a company’s market value by its profit. If accounting changes what counts as profit, the ratio moves without any change in the business or its price. That matters most when comparing today’s valuations with decades of history, because the rules that define “earnings” have not stayed fixed. A 2026 working paper by a Federal Reserve Board economist argues that two such changes explain much of why the Shiller CAPE ratio has looked so high since the 1990s.
Same spending, different earnings
Take two hypothetical companies, each earning $200 million before a $100 million investment and each valued at $3 billion. Company A builds a factory. Accounting capitalizes it and depreciates it over ten years, so this year’s expense is $10 million and reported earnings are $190 million, a P/E of about 15.8. Company B spends the same $100 million on research. Under FASB Statement No. 2, issued in 1974, research and development costs are charged to expense as incurred, so B reports $100 million and a P/E of 30.
Both spent the same cash on something meant to pay off for years. One P/E is nearly double the other only because of where the money went. Scale that up to an index in which software, chip, and drug companies carry growing weight and the whole market’s P/E drifts upward.
How big the research effect is
In Palazzo’s S&P 500 data, R&D rose from 16.0% of adjusted earnings before 1992 to 21.9% afterward. The idea is not new: Campbell and Shiller noted in 2001 that expensing intangible investment biases measured earnings downward, and McGrattan and Prescott estimated that properly measured corporate earnings could be about 27% higher once intangible investment is capitalized. A separate 2026 NBER paper by Atkeson, Heathcote, and Perri finds the aggregate earnings yield falling from 9.5% in 1980 to 4.6% in 2022 while the free-cash-flow yield held near 3.6%, consistent with accounting earnings shrinking relative to cash generation.
Write-downs: large charges that come and go
Special items are the restructuring costs, goodwill impairments, and asset write-downs companies record outside normal operations. Accounting guidance in the mid-1990s, including EITF Issue 94-3 and SFAS No. 121 (1995), standardized when such charges are recognized, and they became far more common: from 1.2% of S&P 500 adjusted earnings before 1992 to 11.7% afterward. They also cluster in recessions. In 2008 aggregate GAAP earnings of the paper’s sample fell 78% while dividends and buybacks fell only 32%. A ten-year average dampens one bad year but does not remove repeated spikes, so they leak into CAPE as well.
Evidence that reported earnings drifted
| Measure | 1957–1979 | 1980–2002 | 2003–2025 |
|---|---|---|---|
| GAAP earnings volatility relative to sales volatility | 1.77× | 3.47× | 7.02× |
| Adjusted earnings volatility relative to sales volatility | 1.53× | 1.60× | 2.55× |
| Average payout ratio, GAAP earnings | 0.54 | 0.80 | 1.09 |
| Average payout ratio, adjusted earnings | 0.48 | 0.52 | 0.66 |
Sales growth was about as volatile in each period, yet GAAP earnings became four times jumpier. And from 2003 to 2025 companies paid out more in dividends and buybacks than they reported earning, on average. That is possible for a while, funded by cash or debt, but its persistence suggests reported earnings understated what companies could sustainably distribute. Against adjusted earnings, the payout ratio stays in a stable range.
Where inflation fits
CAPE already handles one distortion: it converts every month’s price and earnings into today’s dollars with the Consumer Price Index, so a 1960s profit and a 2020s profit are compared in the same purchasing power. Inflation adjustment cannot fix a change in accounting definitions. High inflation also bends earnings in the opposite direction: depreciation based on historical cost and first-in, first-out inventory accounting overstate real profits when prices are rising quickly. That is why the Bureau of Economic Analysis adds inventory valuation and capital consumption adjustments to its national corporate profits, and why some researchers, including Jeremy Siegel, prefer those national-accounts profits for long-run valuation work.
Using earnings-based ratios more carefully
Check which earnings a ratio uses: GAAP, operating, or a company’s own adjusted figure. Company-defined adjusted earnings deserve skepticism, because management can exclude costs that recur every year; the paper’s adjustment instead applies one rule to every firm and does not amortize past R&D, so it is a consistency fix rather than a better estimate of profit. When comparing sectors, remember that a research-heavy company will look more expensive than an asset-heavy one with similar economics. For the market as a whole, use the traditional CAPE as a long-run gauge with this bias in mind. Our Shiller CAPE ratio page tracks it daily, and our guide to EPS, buybacks, and inflation covers another way per-share earnings can mislead.
