How R&D Expensing and Special Items Can Change CAPE's Earnings Base
CAPE's denominator is built from reported earnings. If accounting rules or the types of costs companies incur change, a ten-year average can reflect those changes as well as changes in operating performance.
R&D is expensed as incurred
The FASB's Statement No. 2, issued in October 1974, generally requires research and development costs to be charged to expense when incurred. That treatment reflects the uncertainty of predicting which research projects will produce future economic benefits. It also means an R&D-intensive company reports lower current earnings than it would under a method that capitalized some research costs and amortized them over time.
For example, if a company spends $100 on research in a year, an expense-as-incurred approach records that cost in the period. A capitalization approach would instead record an asset and recognize part of the cost later. The second method requires estimates about useful life, success and amortization; it is not simply a free earnings increase.
Research by Lev and Sougiannis examines how estimated R&D capital relates to accounting values and investor valuations. It supports the idea that R&D measurement can matter, while also illustrating that an adjusted earnings measure depends on assumptions about research assets.
Why a ten-year average may not remove every distortion
A long averaging window reduces the effect of one unusually good or bad year. But it cannot automatically make accounting categories comparable across time. If R&D becomes a larger share of business spending, or if unusual charges become more common, the average can carry those differences forward for years.
Special items present a separate judgment. Restructuring costs or impairments may be unusual for an individual company, yet similar charges can appear repeatedly across the market. Removing every item labelled “special” could overstate normalized earnings if the costs recur or reflect genuine loss of asset value.
The adjustment is a choice, not a correction with no trade-off
Palazzo's working paper argues that these issues affect historical comparisons in CAPE. Its proposed CAPE-H adds R&D expenditures back and excludes special items. The paper frames that as a way to improve consistency in the earnings measure; it does not claim to calculate the economic value of each company's research or identify a uniquely correct earnings number.
When comparing valuation measures, ask what the numerator and denominator include, whether the adjustment is applied consistently across years, and what assumptions are needed. A cleaner time series can be helpful, but the adjustment may also remove costs that investors ultimately bear.
Read CAPE-H explained for the paper's construction, or visit CAPE, Accounting and Long-Term Stock Returns: A Plain-English Pillar Guide for the full guide series. Educational information, not investment advice.