Source: Fama, E. F. & French, K. R. (2008) · Journal of Finance 63(4), 1653–1678
TL;DR
A systematic, skeptical re-examination of the major cross-sectional anomalies — net stock issues, accruals, momentum, profitability, and asset growth — using two complementary tools (value-weighted sorts and Fama–MacBeth cross-section regressions) run separately for tiny, small, and big size groups over July 1963–December 2005. The headline lesson: net stock issues, accruals, and momentum are pervasive (present in all size groups), whereas the asset growth and profitability anomalies are less robust — asset growth is present in tiny/small stocks but absent among big stocks (which hold >90% of market cap), and only the high-profitability side shows up.
The idea
Anomalies are usually documented with equal-weighted (EW) hedge portfolios that can be dominated by tiny stocks (market cap below the 20th NYSE percentile): tiny stocks are only ~3% of total market cap but ~60% of the number of stocks, and they have the widest dispersion in anomaly variables, so they crowd the extreme deciles. Value-weighting fixes that but can be dominated by a few big stocks. Fama and French therefore sort within three size groups (breakpoints at the 20th and 50th NYSE market-cap percentiles) and cross-check with Fama–MacBeth regressions to ask which anomalies are pervasive versus which are microcap phenomena that are hard to trade at scale.
Evidence
Pervasive across all size groups (tiny, small, big): net stock issues, accruals, and momentum — strong in regressions and in sort extremes (with some "chinks in the armor" for net issues and accruals).
Asset growth: anomaly present in tiny and small stocks but absent for big stocks that account for >90% of market capitalization.
Profitability: asymmetric — higher profitability is associated with abnormally higher returns, but there is little evidence that unprofitable firms earn unusually low returns.
Method warning: sorts and regressions can disagree because FM regressions give each tiny stock equal influence (weight by count) while VW sorts do not; regressions also better isolate marginal (unique) effects and functional form.
Why it matters
Locate the anomaly: an effect living only in microcaps is largely untradable at scale — directly relevant to honest, capacity-aware backtesting.
Method discipline: prefer VW evidence, examine size groups separately, and use multiple regression to test which signals carry unique information; this paper is a template for evaluating any new predictor.
Caveats
Conclusions are about pervasiveness and tradeability, not about why the anomalies exist (risk vs mispricing).
Sample ends in 2005; relations may have changed out of sample.
Both tools have weaknesses (sorts hide marginal effects/functional form; regressions over-weight tiny stocks) — the paper's point is to triangulate, not to trust either alone.
Key references
Fama, E. & French, K. (2008) — Dissecting Anomalies — Journal of Finance
Fama, E. & MacBeth, J. (1973) — Risk, Return, and Equilibrium — JPE
Sloan, R. (1996) — Do Stock Prices Fully Reflect Accruals? — Accounting Review
Cooper, Gulen & Schill (2008) — Asset Growth and the Cross-Section of Stock Returns — Journal of Finance
Provenance: verified/generated from the paper's full text.
