Small-capitalization stocks earned higher risk-adjusted returns than large-cap stocks over 1936–1975 — the first documented size effect. The relation is strongly nonlinear, concentrated in the very smallest firms, and is not explained by the CAPM. This paper seeded the size factor (later SMB).
Direction:negative — small firms earn higher average and risk-adjusted returns than large firms.
Shape: highly nonlinear. The premium is concentrated in the smallest firms; there is little difference between mid- and large-caps. This nonlinearity matters for both interpretation and implementation.
OSAP predictor: Size.
How to construct it
Sorting variable: market equity at formation.
Universe: NYSE common stocks in the original study (1936–1975); later work adds AMEX/NASDAQ, which intensifies the apparent effect because of tiny names.
Portfolio formation: rank by market cap; the modern SMB convention reforms annually in June using NYSE breakpoints.
Long / short: long small-cap, short large-cap.
Weighting: value-weighting within legs is standard (SMB); equal-weighting massively amplifies the historical effect via microcaps.
Evidence and replication
Period
Sharpe (approx)
Ann. Return
T-stat
Source
IS (1936–1975)
modest
economically meaningful small-cap premium
significant
this paper
OOS (post-1981)
~0
largely vanished
—
post-publication
OSAP replication (Size)
clear IS, weak OOS
—
—
Chen & Zimmermann 2022
Banz documented a sizeable small-firm premium not explained by beta, but cautioned that it was unknown whether size was a true risk factor or a symptom of model misspecification.
The size effect is among the most-decayed classic anomalies: it weakened almost immediately after publication and is widely considered close to zero net of microcaps, the January seasonal, and delisting/survivorship biases that inflated early estimates.
It survives mainly as a conditioning variable — size interacts with value, quality, and illiquidity — rather than as a standalone premium (see Asness et al., "Size Matters, If You Control Your Junk").
Why it might work
Risk-based: small firms may load on distress, illiquidity, or funding-constraint risks not captured by market beta. The premium would then be compensation for those exposures.
Liquidity: small stocks are illiquid; much of the "size" premium overlaps the illiquidity premium (Amihud 2002).
Data artifacts: a meaningful portion of the early estimate reflects survivorship and delisting biases in microcap data, plus the January effect — i.e., not a robust economic premium.
Quality conditioning: Asness, Frazzini, Israel, Moskowitz & Pedersen (2018) show a clean size premium re-emerges once you control for junk/quality — small quality firms are rewarded; small junk firms are not.
Limitations and risks
Decay: the headline standalone premium essentially disappeared after publication.
Microcap and January dependence: what remains lives in tiny, illiquid stocks and a calendar seasonal — minimal scalable capacity.
Data bias: early results overstated by survivorship/delisting issues.
Only useful conditioned: practically valuable mainly in combination with quality/value, not alone.
No free full text: original is paywalled; see DOI.
Key references
Banz, R. (1981) — The Relationship Between Return and Market Value of Common Stocks — Journal of Financial Economics — DOI: 10.1016/0304-405X(81)90018-0
Fama, E. & French, K. (1992) — The Cross-Section of Expected Stock Returns — Journal of Finance
Asness, C., Frazzini, A., Israel, R., Moskowitz, T. & Pedersen, L. (2018) — Size Matters, If You Control Your Junk — Journal of Financial Economics
Amihud, Y. (2002) — Illiquidity and Stock Returns — Journal of Financial Markets
Chen, A. & Zimmermann, T. (2022) — Open Source Cross-Sectional Asset Pricing — Critical Finance Review
