Factor Investing

The Fama-French Three-Factor Model: Size, Value, and Market Risk

Summary by Robert Gorak · Published June 9, 2026 · Last reviewed June 9, 2026

Eugene F. Fama and Kenneth R. French·1993·Journal of Financial Economics
Sample: 25 stock portfolios formed on size and BE/ME quintiles; 7 bond portfolios; 342 monthsData: CRSP / COMPUSTATPeriod: July 1963 to December 1991

The Fama-French three-factor model uses RM-RF, SMB (small minus big), and HML (high minus low BE/ME) to price assets. Fama and French (1993) analyzed NYSE, Amex, and NASDAQ stocks from July 1963 to December 1991. The paper appeared in the Journal of Financial Economics under the title "Common Risk Factors in the Returns on Stocks and Bonds". Adding SMB and HML raised R² from 0.61–0.70 to 0.94–0.97 for the smallest-size quintile portfolios. The lowest three-factor R² across all 25 portfolios was 0.83.

What the Study Found

Average excess returns on the 25 size- and BE/ME-sorted stock portfolios ranged from 0.32% to 1.05% per month across the 29-year sample. SMB produced an average return of 0.27% per month (t = 1.73). HML produced an average return of 0.40% per month (t = 2.91). The average excess market return (RM-RF) was 0.43% per month. Bond portfolio excess returns were all less than 0.15% per month. Only one of seven exceeded 1.5 standard errors from 0.

Methodology

Fama and French drew stock data from CRSP and accounting data from COMPUSTAT for NYSE, Amex, and (after 1972) NASDAQ stocks. The sample covers July 1963 to December 1991, yielding 342 monthly observations across 25 size-BE/ME portfolios and 7 bond portfolios. The study used time-series regressions following Black, Jensen, and Scholes (1972), regressing excess portfolio returns on five mimicking-factor returns. Controls included the value-weighted market return, NYSE size median breakpoints, and COMPUSTAT book equity computed net of deferred taxes and preferred stock.

Key Statistics

Metric Finding Context
Three-factor R² — smallest-size portfolios 0.94 to 0.97 Up from 0.61–0.70 with market factor alone
Minimum three-factor R² (all 25 stock portfolios) 0.83 Largest-size, highest-BE/ME portfolio
Average HML monthly return 0.40% per month (t = 2.91) July 1963–December 1991
Average SMB monthly return 0.27% per month (t = 1.73) July 1963–December 1991
Average RM-RF monthly return 0.43% per month About 5% per year
SMB–HML correlation −0.08 Monthly, 1963–1991
Average excess return range (25 stock portfolios) 0.32% to 1.05% per month July 1963–December 1991
Average bond portfolio excess return Less than 0.15% per month All government and corporate bond portfolios
Three-factor model R(t) − RF(t) = a + b[RM(t) − RF(t)] + s·SMB(t) + h·HML(t) + e(t) Tests whether intercepts are indistinguishable from 0
SMB construction SMB = ⅓(S/L + S/M + S/H) − ⅓(B/L + B/M + B/H) Mimicking portfolio for size risk factor
HML construction HML = ½(S/H + B/H) − ½(S/L + B/L) Mimicking portfolio for BE/ME risk factor

Why This Matters

The three-factor model gives portfolio managers a framework to decompose return sources beyond market beta alone. Allocating to small-cap or high-BE/ME stocks loads on SMB and HML, which the model interprets as compensation for common risk factors. Institutional strategies that explicitly target size or value tilts rest on this decomposition. The bond-market results establish that stock and bond returns share variation through two term-structure factors, linking equity and fixed-income risk pricing in a single model.

Frequently Asked Questions

The three-factor model raises R² to 0.94–0.97 for the smallest-size quintile portfolios, up from 0.61–0.70 with the market factor alone. The minimum three-factor R² across all 25 size-BE/ME portfolios was 0.83. SMB t-statistics on stock portfolio slopes were greater than 4 for all but one of the 25 portfolios.

SMB averaged 0.27% per month and HML averaged 0.40% per month (t = 2.91) from July 1963 to December 1991. The two factors had a correlation of −0.08. Both are constructed from six value-weighted portfolios formed on size and BE/ME sorts of NYSE, Amex, and NASDAQ stocks.

Bond portfolio excess returns (government and corporate) were all less than 0.15% per month. Only one of seven exceeded 1.5 standard errors from 0. TERM and DEF captured most bond return variation. Their average returns were 0.06% and 0.02% per month; Fama and French (1993) concluded they explain almost none of average excess stock returns.

Each June from 1963 to 1991, NYSE, Amex, and NASDAQ stocks were sorted into five size quintiles using NYSE median breakpoints and five BE/ME quintiles. The 25 intersections yielded value-weighted portfolios held monthly from July of year t to June of year t + 1. Returns from July 1963 to December 1991 — 342 months — served as the dependent variables in the time-series regressions.

Source

Eugene F. Fama and Kenneth R. French (1993). Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics.

Read the full paper