Source: Lakonishok, J., Shleifer, A. & Vishny, R. W. (1994) · Journal of Finance 49(5), 1541–1578 · NBER WP 4360 (1993)
TL;DR
Value strategies (buy stocks cheap relative to fundamentals) beat glamour strategies by roughly 8% per year, and the paper argues this premium comes from mistakes of the typical investor — extrapolating past growth too far — not from extra fundamental risk. Naive investors over-extrapolate glamour firms' strong past growth and shun poor past performers, so value stocks become underpriced; the abnormal returns persist for years and value does not underperform in bad states of the world.
What anomaly it documents
Predictor: value vs. glamour, measured by low price multiples (book-to-market, cash-flow-to-price, earnings-to-price) and/or low past growth in sales (GS). Best classifications combine an expected-performance measure (e.g. C/P or E/P) with a past-performance measure (GS).
Direction: positive — high B/M, high C/P, high E/P, and low past-sales-growth stocks earn higher subsequent returns.
Shape: roughly monotonic across deciles; the spread is ~7–8% per year for one-way sorts and larger for two-way sorts.
OSAP predictor family: value (BM, CFtoP, EPtoP) and sales growth.
How to construct it
Universe: NYSE and AMEX common stocks (requires 5 years of past accounting data).
Sample: end-April 1963 to end-April 1990; portfolios formed annually starting end-April 1968 and tracked up to 5 years after formation.
Sorts: annual deciles on the chosen ratio (B/M, C/P, E/P) or on past 5-year sales growth; also two-dimensional independent sorts (e.g. C/P × GS).
Weighting: equal-weighted within portfolios; annual buy-and-hold with annual rebalancing.
Long / short: long value (cheap / low past growth), short glamour (expensive / high past growth).
Evidence and replication
Period
Notes
Source
1968–1990 (formation), held 5 yrs
One-way value premium ~7.8%/yr; B/M decile spread sizable
this paper
Two-way C/P × GS
best classification: ~10–11%/yr in year-1 returns; ~8.7%/yr average glamour-vs-value spread over post-formation years
this paper
5-year cumulative
difference in size-adjusted returns ~22.7% over five years
this paper
Robustness
premium does not shrink restricting to the largest 50% or 20% of stocks
this paper
Why it might work
Extrapolation / expectational errors: investors over-extrapolate glamour firms' high past growth (implied future growth far exceeds what is realized — e.g. glamour earnings expected to grow ~11% faster but grow slower), so glamour is overpriced and value underpriced.
Behavioral, not risk-based: the authors find little support for the risk story — value has comparable or lower beta and standard deviation, and crucially does particularly well in extreme down markets and recessions (the 1973–75 recession), the opposite of what a risk premium for bad-state losses would predict.
Limitations and risks
Behavioral interpretation is contested; Fama & French (1992, 1993) attribute the value premium to compensation for systematic distress risk.
Equal-weighted, annual-rebalance design with a pre-1990 sample; transaction costs and small-stock tilts can erode realized returns.
The value premium has weakened and decayed substantially in subsequent decades (post-publication; cf. McLean & Pontiff 2016).
Key references
Lakonishok, J., Shleifer, A. & Vishny, R. (1994) — Contrarian Investment, Extrapolation, and Risk — Journal of Finance
Fama, E. & French, K. (1992) — The Cross-Section of Expected Stock Returns — Journal of Finance
De Bondt, W. & Thaler, R. (1985) — Does the Stock Market Overreact? — Journal of Finance
Provenance: verified/generated from the paper's full text.
Reference replication on ConvexPi
An open, verified replication of this strategy is maintained at convexpi/replications. It recomputes the strategy from underlying building blocks and scores it out of sample (the McLean & Pontiff test):
