Source: Fama & French (1992) · The Journal of Finance · DOI: 10.1111/j.1540-6261.1992.tb04398.x
TL;DR
Two simple firm characteristics — size (market equity) and book-to-market equity (BE/ME) — capture the cross-sectional variation in average US stock returns over 1963–1990, while market beta has essentially no marginal explanatory power. This paper is the empirical foundation of the size and value factors and the proximate cause of death for the one-factor CAPM.
What anomaly it documents
The paper is a horse race among the candidate variables that prior literature linked to average returns: market beta, size, E/P, leverage, and book-to-market. Run jointly in cross-sectional regressions, the result is stark:
Size (ME): negative relation — small stocks earn higher average returns than large stocks (the size premium).
Book-to-market (BE/ME): strong positive relation — high-BE/ME ("value") stocks earn higher average returns than low-BE/ME ("growth") stocks. This is the stronger of the two effects.
Market beta: once size is controlled for, the relation between beta and average return is flat — economically and statistically indistinguishable from zero over 1963–1990. This directly contradicts the CAPM.
E/P and leverage: their explanatory power is absorbed by the size–BE/ME combination, so they add nothing once those two are included.
The relevant OSAP predictors are size and book-to-market (the linked signal here is BookLeverage, one of the leverage variables the paper shows is subsumed by BE/ME). Size and value are both classified as clear, replicable anomalies in the Chen-Zimmermann database, though both have weakened materially out of sample (see below).
How to construct it
The value (BE/ME) factor — the paper's headline result, formalized as HML in Fama & French (1993):
Sorting variable: book equity divided by market equity. Book equity from the fiscal year ending in calendar year t−1; market equity from the end of December of t−1. A 6-month-plus gap ensures accounting data are publicly known before formation.
Universe: NYSE + AMEX + NASDAQ, nonfinancial firms (financials excluded because high leverage means something different for them). Require positive book equity.
Breakpoints: NYSE breakpoints (to avoid the tiny-NASDAQ-stock distortion), even though the universe includes AMEX/NASDAQ.
Portfolio formation: at the end of June of year t, sort stocks on BE/ME.
Long leg / short leg: long high-BE/ME (value), short low-BE/ME (growth). The standard HML uses a 2×3 double sort on size and BE/ME and averages the value-minus-growth return across the two size buckets.
Weighting: value-weighted within portfolios.
Rebalancing: annually, each June.
The size (SMB) factor is built analogously: small-minus-big, value-weighted, NYSE breakpoints, reformed each June.
Evidence and replication
Period
Sharpe (approx)
Ann. Return
T-stat
Source
IS value (BE/ME), 1963–1990
~0.5
high-minus-low spread ≈ 0.5%/mo+
~3+
this paper
IS size (ME), 1963–1990
~0.2–0.3
≈0.2–0.3%/mo
~2–3
this paper
Market beta (controlling for size)
~0
flat
≈0
this paper
OOS value (post-1990)
lower
much weaker, near-zero post-2007
—
post-publication
OOS size (post-1981 already weak)
~0
largely vanished
—
post-publication
OSAP replication
clear, positive IS
—
—
Chen & Zimmermann 2022
Key points:
The central empirical claim — a flat beta/return relation — was so influential it is often summarized as "beta is dead." Subsequent work (e.g., critiques about measurement and conditioning) nuances this, but the marginal-beta result held up in the sample.
Value (BE/ME) is the more durable of the two effects in-sample, with the largest spread among the variables tested.
Out of sample, both factors decayed sharply. The size premium was already fragile after its 1981 publication (Banz) and is widely considered close to zero net of microcap/illiquidity effects. The value premium delivered for decades but suffered a prolonged, severe drawdown from roughly 2007–2020, prompting active debate about whether it is impaired (intangibles mismeasuring book value) or merely cyclical.
This is squarely a McLean-Pontiff case study: heavily published, heavily traded, and materially weaker post-publication.
Why it might work
Risk-based (Fama & French's own view): size and value proxy for exposure to distress or fundamental risk not captured by the market. Value firms are distressed, have high operating leverage, and do badly in bad times — so their premium is compensation for systematic risk. This view motivated the three-factor model.
Behavioral (Lakonishok, Shleifer & Vishny 1994): value works because investors extrapolate past growth too far, overpricing glamour stocks and underpricing value stocks; the premium is correction of mispricing, not risk compensation.
Intangibles / measurement: a modern strand argues book value has become a poor measure of capital as intangibles (R&D, brand) grew, mechanically distorting BE/ME and partly explaining value's post-2007 struggles.
The risk-vs-mispricing debate over value is one of the longest-running in asset pricing and remains unsettled.
Limitations and risks
Out-of-sample decay: both factors are far weaker post-publication; the size premium in particular is close to nonexistent net of microcap and January effects.
Value's lost decade: the 2007–2020 drawdown shows the premium can disappear for a decade-plus — a real risk for anyone allocating to it.
Microcap dependence (size): much of the historical size effect lives in tiny, illiquid stocks that are expensive or impossible to trade at scale.
Look-ahead and survivorship: correct implementation requires point-in-time book equity and the June-formation lag; naive backtests that use restated or contemporaneous accounting data overstate returns.
Crowding: as foundational, indexed factors, both are heavily harvested, compressing the premium.
Definition sensitivity: results depend on the BE/ME definition (treatment of negative book equity, deferred taxes, financials exclusion) and NYSE breakpoints.
Key references
Fama, E. & French, K. (1992) — The Cross-Section of Expected Stock Returns — Journal of Finance — DOI: 10.1111/j.1540-6261.1992.tb04398.x
Fama, E. & French, K. (1993) — Common Risk Factors in the Returns on Stocks and Bonds — Journal of Financial Economics
Banz, R. (1981) — The Relationship Between Return and Market Value of Common Stocks — Journal of Financial Economics
Lakonishok, J., Shleifer, A. & Vishny, R. (1994) — Contrarian Investment, Extrapolation, and Risk — Journal of Finance
Fama, E. & French, K. (2015) — A Five-Factor Asset Pricing Model — Journal of Financial Economics
McLean, R. D. & Pontiff, J. (2016) — Does Academic Research Destroy Stock Return Predictability? — Journal of Finance
Chen, A. & Zimmermann, T. (2022) — Open Source Cross-Sectional Asset Pricing — Critical Finance Review
