The size effect is the tendency for small-capitalization stocks to earn higher average returns than large-cap stocks, after controlling for market risk. First documented by Banz (1981), who found that NYSE small firms earned higher risk-adjusted returns over at least 40 years, with the effect concentrated in the very smallest firms and non-linear in market value.
Construction
SMB (Small Minus Big) is the difference between the average return on small-stock portfolios and big-stock portfolios, using the NYSE median market capitalization as the size breakpoint. In the Fama-French three-factor model, SMB is constructed to be orthogonal to ME by averaging across book-to-market groups.
Key empirical evidence
Fama and French (1992) show that average monthly returns range from 1.64% (smallest decile) to 0.90% (largest decile). The FM regression slope on ln(ME) is -0.15%/month (t=-2.58). Size remains significant after controlling for book-to-market.
Fama and French (1993) report SMB averages 0.27%/month (t=1.73) in the 1963-1991 sample. In the five-factor model (1963-2013), SMB averages 0.29%/month.
Weaker than value
Size has a consistently weaker role than value in explaining cross-sectional returns. The size spread (0.74%/month) is less than half the ME spread (1.53%/month). The Asness et al. (2014) comparison across factors shows SMB has the lowest Sharpe ratio (0.26) among the standard factors, compared to 0.39 for HML and 0.50 for UMD.
Interaction with profitability and investment
The five-factor model reveals that the problematic portfolios are specifically small stocks that invest aggressively despite low profitability. These produce large negative intercepts (-0.47%/month, t=-5.89) that drive model rejection. Large stocks with similar characteristics do not pose this problem.
Mechanical critique (Berk 1995)
Berk (1995) argues that size-return correlations are mechanical rather than anomalous. Market capitalization equals the present value of expected cash flows discounted at the expected return. Holding cash flows constant, a higher expected return implies a lower market cap. Size sorts therefore proxy for expected returns by construction: ranking stocks by market cap is effectively ranking them by their discount rates. The “anomaly” dissolves once one recognizes that any variable containing price in the denominator will be correlated with expected returns. This critique applies broadly to anomalies using market-price-scaled variables (E/P, B/M, D/P) but is most direct for size, where market cap is the sorting variable itself.
Migration mechanism (Fama and French 2007)
Fama and French (2007) decompose the size premium into contributions from stocks that stay in the same portfolio versus those that migrate across size and value groups. The size premium (1927-2006) traces almost entirely to the 8-12% of small-cap market capitalization that earns extreme positive returns and migrates to a big-cap portfolio each year. These migrating stocks earn average excess returns exceeding 60% per year. Small-cap stocks that remain small or deteriorate actually lean against the premium. The size premium is essentially a lottery-ticket effect: most small stocks underperform, but a few spectacular winners more than compensate. See migration for the full framework.
No size effect after beta adjustment (Asness 2020)
Asness (2020) argues there is no “simple” size effect (returns to small versus large after adjusting for market-beta). Two problems eliminate the historical premium:
- Database corrections: original databases overstated small-stock returns through biased delisting return estimates. Re-running original tests on corrected data over the same periods produces a smaller premium.
- Beta misestimation from illiquidity: small stocks trade infrequently, so contemporaneous regressions underestimate their true market beta. Adding lagged market returns to regressions reveals that Decile 1 (smallest) stocks have a total beta of approximately 1.35 (monthly) or 1.19 (daily), far above the 0.74 contemporaneous daily beta. After this correction, the alpha is essentially zero or slightly negative.
Daily data makes this especially stark: without lag adjustment, the smallest decile appears to have a beta of 0.74 (below 1.0), which is obviously wrong. Adding one-day and 2-10 day lagged market returns restores the beta to 1.19 and eliminates the alpha entirely.
However, Asness separately shows that after controlling for quality (the “junk” factor), small stocks look much more impressive. Small stocks are far lower quality than large stocks, so even matching large-stock returns is a feat given this headwind. The quality-adjusted size effect is significant and robust, but this is distinct from the traditional “simple” size effect. See size-effect-debate for the full controversy.
Structural decline of the size premium (BCA Research 2024)
BCA Research (2024) argues the small-cap premium is disappearing for structural reasons beyond statistical adjustments. Building on the migration framework, they document that migration rates from small to large caps have declined to roughly half their level at the turn of the century. The causes:
- Venture capital growth: the best potential small-cap companies stay private longer, backed by VC funding, and either never go public or enter public markets already at large-cap scale. Late-stage VC deal valuations now routinely exceed the Russell 1000/2000 breakpoint.
- Big tech acquisitions: successful small companies are acquired by large incumbents before they can grow organically into large-cap status. VC-backed exits via acquisition have surged.
- Declining IPOs: the number of US publicly listed companies has collapsed since 2000, from roughly 8,000 to under 4,000, while high-income countries excluding the US have not seen the same decline.
- Junkification of small caps: with the best companies removed, small-cap indices have become increasingly populated by lower-quality firms. The quality of new small-cap entrants has deteriorated.
BCA concludes that any future small-cap outperformance will be cyclical rather than structural, and advises against a passive structural allocation to US small-cap indices.
Role in factor models
SMB appears in every major multi-factor model as a control variable. Its theoretical motivation is weaker than value or momentum — it may proxy for liquidity risk, information asymmetry, or exposure to macroeconomic shocks that disproportionately affect small firms.
Sources
- The Cross-Section of Expected Stock Returns (File, DOI)
- Common risk factors in the returns on stocks and bonds (File, DOI)
- A five-factor asset pricing model (File, DOI)
- A Critique of Size-Related Anomalies (File, DOI)
- Migration (File, DOI)
- The Great Small Cap Heist (File)
- There Is No Size Effect: Daily Edition (File, URL)
- The Relationship Between Return and Market Value of Common Stocks (File, DOI)