The existence and nature of the size premium has been debated since Banz (1981) first documented that small-cap stocks earn higher average returns. Multiple lines of critique have progressively weakened the case for a simple, unconditional size effect.
The mechanical critique (Berk 1995)
Berk (1995) provides the earliest fundamental challenge. Market capitalization is the present value of expected future cash flows discounted at the expected return. Sorting stocks by market cap therefore sorts them by discount rates by construction: smaller market cap mechanically implies higher expected returns, holding cash flows constant. The size “anomaly” is not that small stocks earn more, but rather that a variable containing price in the denominator predicts returns, which is tautological. This critique extends to any market-price-scaled sorting variable (B/M, E/P, D/P), but applies most directly to size where market cap is the sort variable itself.
Database and beta problems (Asness 2020)
Asness (2020) identifies two empirical problems that eliminate the historical size premium:
-
Delisting bias: original databases (used in Banz 1981 and related work) overstated small-stock returns through excessively optimistic delisting return estimates. Because small firms delist more frequently, the bias disproportionately inflates their measured returns. Corrected databases over the same sample periods yield a smaller premium.
-
Beta misestimation from illiquidity: small stocks trade infrequently. Contemporaneous market regressions underestimate their true beta because price changes reflect past market moves that were never recorded. Adding lagged market returns reveals:
- Monthly data: Decile 1 total beta rises from 1.10 (contemporaneous) to 1.35 (with one-month lag), eliminating the alpha.
- Daily data: Decile 1 contemporaneous beta is 0.74 (obviously wrong), but adding 1-day and 2-10 day lags restores it to 1.19. The alpha flips from +2.3% to -0.6% per annum.
After these corrections, the “simple” size effect (small minus large, net of market beta) is zero.
The quality/junk adjustment
In separate work (“Size Matters, If You Control Your Junk,” 2015), Asness and co-authors show that small stocks are far lower quality than large stocks. Since quality is a rewarded factor, small stocks face a severe headwind. The fact that small stocks approximately match large stocks’ beta-adjusted returns despite this headwind is itself impressive. After controlling for quality, the size effect becomes significant and robust.
This creates an apparent contradiction that Asness addresses directly: both findings are consistent. The simple size effect (net of market beta only) does not exist. The quality-adjusted size effect (net of market beta and quality) is strong. These are different claims about different things.
Migration dynamics (Fama and French 2007)
Fama and French (2007) show that whatever size premium exists is driven by a specific mechanism: the 8-12% of small-cap market capitalization that earns extreme positive returns (60%+ per year) and migrates to big-cap status. Most small stocks underperform large stocks. The premium is a lottery-ticket effect concentrated in a few spectacular winners each year. See migration.
Structural VC thesis (BCA Research 2024)
BCA Research (2024) extends the migration framework to argue the size premium is structurally disappearing. Migration rates from small to large caps have fallen to roughly half their level at 2000. The cause is the growth of venture capital and big tech acquisitions:
- The best potential small-cap companies stay private longer (VC-funded) or are acquired before reaching public markets.
- Late-stage VC deal valuations now exceed the Russell 1000/2000 breakpoint, so companies that do IPO often enter as large caps directly.
- The number of US public companies has roughly halved since 2000.
- Small-cap indices have become the “market’s junkyard,” populated disproportionately by lower-quality firms.
BCA concludes that future small-cap outperformance, if any, will be cyclical rather than structural.
Synthesis
The various critiques attack different aspects of the size effect:
| Critique | Type | Implication |
|---|---|---|
| Berk (1995) | Theoretical | Size-return link is mechanical, not an anomaly |
| Database corrections | Empirical | Original premium was overstated |
| Beta misestimation | Empirical | True beta adjustment eliminates alpha |
| Quality/junk | Empirical | Size works only after quality control; different from original claim |
| Migration | Mechanism | Premium driven by rare outlier transitions |
| VC/structural | Market structure | Migration channel is drying up |
The cumulative weight of evidence suggests the simple, unconditional size premium is at best fragile and at worst nonexistent. What remains is a more nuanced finding: small stocks outperform on a quality-adjusted basis, but this is not easily implementable and the structural conditions that historically generated even the raw premium may be eroding.