The size factor is the empirical tendency for small-capitalization stocks to earn higher average returns than large-capitalization stocks over long horizons. First documented by Rolf Banz in 1981 and formalized as the "SMB" (Small Minus Big) leg of the Fama-French three-factor model in 1993, it is one of the original equity "anomalies" and a cornerstone of factor investing. Its core tension is that, unlike value or momentum, the standalone size premium has proven fragile out of sample — it has been the most contested of the classic factors, and much of the academic case for it now rests on combining size with a quality control rather than buying small caps indiscriminately.

How it's calculated / formed

SMB is the textbook way the factor is measured. In the Fama-French methodology, stocks are independently double-sorted each year:

• Size split: the universe is divided into two groups — Small and Big — using the NYSE median market cap as the breakpoint (so micro-caps don't dominate the count).
• Book-to-market split: independently sorted into three groups using the NYSE 30th and 70th percentiles (Growth / Neutral / Value).

This 2x3 grid yields six value-weighted portfolios. SMB is the average return of the three small portfolios minus the average return of the three big portfolios:

SMB = ⅓(Small Value + Small Neutral + Small Growth) − ⅓(Big Value + Big Neutral + Big Growth)

Averaging across the three book-to-market buckets is deliberate: it strips the value tilt out of SMB so it isolates size. (Ken French's data library publishes the daily/monthly SMB series, with a slightly modified construction in the 5-factor model that also neutralizes profitability and investment.)

In a regression context, a portfolio's SMB beta is the slope from regressing its excess returns on the SMB series. A positive SMB beta means small-cap exposure; a negative beta means the portfolio behaves like large caps. "Size" here means market cap, not revenue, headcount, or book value.

How it's used in practice

• Performance attribution. This is arguably SMB's most durable real-world use. Running a fund's returns against the three- or five-factor model decomposes its track record: a manager who is simply long small caps will show a high SMB beta, revealing that "alpha" as a known factor tilt rather than genuine skill.
• Cost-of-capital / valuation. Practitioners add a size premium to discount rates for small private or thinly traded firms (the Ibbotson/Duff & Phelps "size premia" tables institutionalized this), reflecting that small firms are riskier and less liquid.
• Strategic factor tilts. Investors deliberately overweight small caps via index funds, dedicated small-cap factor ETFs, or multi-factor products. Crucially, sophisticated implementations rarely buy raw size — they combine it with quality, value, or profitability screens (see below).
• Diversification within a factor sleeve. Because the size premium's timing differs from value and momentum, it is held as one leg of a diversified multi-factor portfolio rather than a standalone bet.

Adoption, debate & evidence

The size factor is the most heavily disputed of the original anomalies, and the honest picture is "folklore says small beats big; the measured record says only conditionally."

• The premium shrank dramatically post-publication. Banz's original work (data roughly 1936–1975, with a full sample back to 1926) reported a sizable small-cap edge. But research recapping the period since publication finds the monthly premium fell to a fraction of its original size — one widely cited decomposition (Advisor Perspectives / Larry Swedroe, citing Fama-French data) puts the U.S. monthly size premium at roughly 0.30% over the long historical sample (≈1926–1981) versus about 0.08% from 1982 through 2017, i.e. economically and statistically marginal out of sample.
• Data-quality doubts about the original finding. Later reconstructions suggest delisting and survivorship errors in early datasets inflated the original Banz result; some find the premium was not even statistically significant over its own discovery window once corrected.
• The January / micro-cap concentration. A large share of the historical premium clustered in January and in the smallest, least-liquid micro-caps — returns that are hard or costly to actually capture after transaction costs and spreads.
• The "control your junk" resurrection. The most influential rehabilitation is Asness, Frazzini, Israel, Moskowitz & Pedersen, Size Matters, If You Control Your Junk (2018, Journal of Financial Economics). They argue raw SMB is weak because small caps are disproportionately low-quality ("junk") firms, which drag returns. After controlling for quality (the QMJ factor), they report a size premium that is stronger, more stable across time, more consistent across seasons and across 24 international markets and ~30 industries, and not concentrated in micro-caps — on par with value and momentum. This is now the mainstream institutional defense of size: it is a real premium, but only cleanly visible when you avoid junk.

Net: the size factor is widely used (attribution, cost of capital, factor tilts) but its standalone return edge is genuinely contested. The defensible claim is conditional, not absolute.

Strengths & limitations

When it works / is useful: as an attribution and risk-modeling tool it is essentially indispensable — SMB beta is a standard lens for understanding any equity portfolio. As a return driver, the evidence is strongest when size is combined with quality/profitability and implemented patiently in liquid small caps rather than micro-caps.

When it fails: raw small-cap tilts have delivered long, painful stretches of underperformance versus large caps (notably much of the 2010s mega-cap-led market). The premium is regime-dependent, illiquidity-tainted, and vulnerable to transaction costs precisely where it looks largest (micro-caps).

The single most common misuse: treating "small caps outperform" as a reliable standalone law and buying the smallest, cheapest names — which loads up on distressed, unprofitable junk and often destroys the premium it was meant to harvest. The corollary misuse is double-counting: a manager whose "alpha" is just a passive small-cap tilt.

Sources

• Banz, R. (1981), The Relationship Between Return and Market Value of Common Stocks — origin of the size effect.
• Fama, E. & French, K. (1993), three-factor model; Ken French Data Library — SMB construction (2x3 size/BtM sort) and series.
• Asness, Frazzini, Israel, Moskowitz & Pedersen (2018), Size Matters, If You Control Your Junk, J. of Financial Economics — quality-controlled size premium. (AQR / Jacobs Levy Center working paper.)
• Advisor Perspectives / L. Swedroe, "Has the Size Premium Disappeared?" — post-publication shrinkage, January effect, data-error critique.
• Corporate Finance Institute; arXiv 1708.00644 ("The 'Size Premium' in Equity Markets: Where is the Risk?") — SMB definition and contested-risk discussion.
• Dispute flagged: the standalone size premium's existence/significance is genuinely contested; only the quality-controlled version has broad academic support.