Key takeaways
– HML (High Minus Low) is the “value premium”: the return spread between high book-to-market (value) and low book-to-market (growth) stocks.
– HML is one of the original three factors in the Fama–French three-factor model and remains in the five-factor expansion.
– You can construct HML from market data (book equity and market equity), then use time-series regression to estimate a portfolio’s HML beta and how much of its returns come from value exposure.
– Interpreting HML requires care: factor definitions, weighting, look-ahead bias, sample period, and transaction costs all affect results.
Understanding HML (the concept)
– What it measures: HML = average return of high book-to-market portfolios − average return of low book-to-market portfolios. It captures the empirical tendency, over many historical samples, for “value” stocks (high book-to-market) to outperform “growth” stocks (low book-to-market).
– Why it matters: In the Fama–French framework, HML is a systematic source of cross-sectional expected returns. When a portfolio has positive HML exposure, part of its historical excess return can be attributed to taking value risk rather than manager skill.
Classic academic background
– The Fama–French three-factor model (1993) augmented CAPM by adding two factors: size (SMB, Small Minus Big) and value (HML). The model better explains cross-sectional returns than CAPM in many tests. (See Fama & French, JFE 1993.)
– In 2015 Fama and French expanded the framework to five factors by adding profitability (RMW) and investment (CMA) while retaining HML as the value factor. (See Fama & French, JFE 2015.)
– Practical and empirical performance of the model can vary by market, sample period, and portfolio construction method (e.g., Dhaka Stock Exchange study cited by Investopedia).
How HML is constructed — step-by-step (practical)
Note: many practitioners use data vendors (CRSP/Compustat, Bloomberg, FactSet) rather than assembling raw data from public filings. Below is the classic approach used to replicate Fama–French factors.
1. Select the cross-section and data frequency
• Universe: e.g., all common stocks on an exchange or broad market index.
• Frequency: monthly returns are common for factor construction. Use consistent return and accounting periods.
2. Compute book-to-market (B/M) ratio
• Book equity = (total shareholders’ equity) adjusted for preferred stock and deferred taxes (follow consistent definition).
• Market equity (market capitalization) = price × shares outstanding at the portfolio formation date.
• B/M = book equity / market equity.
3. Form portfolios by size and B/M
• Classic Fama–French: split stocks into two size groups (Small, Big) at the median market equity, and into three B/M groups (Low, Medium, High) at 30th and 70th percentiles (or tertiles). This yields six portfolios: S/H, S/M, S/L, B/H, B/M, B/L.
• Alternatives: quintiles or deciles for more granularity; value-weighted or equal-weighted portfolios.
4. Calculate portfolio returns
• Compute monthly return for each of the 6 portfolios over the next period (after formation). Use the same weighting method chosen (value- or equal-weighted). Account for delistings and dividends.
5. Construct HML
• Classic Fama–French formula: HML = 0.5*(R_S,H + R_B,H) − 0.5*(R_S,L + R_B,L)
(i.e., average of the High B/M portfolios across sizes minus average of the Low B/M portfolios across sizes).
• You can also compute an overall high-minus-low across all sizes (depending on your methodology).
6. Repeat monthly to build a time series of HML returns.
Estimating HML beta and attribution (regression steps)
1. Prepare excess returns:
• For a portfolio (or asset) i, compute excess returns: Ri,t − Rf,t, where Rf is the risk-free rate.
2. Run time-series OLS regression (three-factor model):
• Model: Ri,t − Rf,t = αi + βi,Mkt*(Rm,t − Rf,t) + βi,SMB*SMB_t + βi,HML*HML_t + εi,t
• For the five-factor model add RMW and CMA as explanatory factors.
3. Interpret coefficients:
• βi,HML (HML beta): sensitivity of portfolio i to the value factor.
• β > 0: portfolio has value tilt (exposed positively to HML).
• β < 0: portfolio tilts toward growth (negative exposure).
• Size of β: how many units of HML exposure per unit of factor movement (e.g., β = 0.5 means half as sensitive as a unit HML).
• α (alpha): the average abnormal return unexplained by the model. If statistically different from zero, it may indicate manager skill or model misspecification.
Practical implementation checklist (analyst-friendly)
– Data quality: use reliable market-cap and accounting data, handle cross-listings/delisted stocks, and avoid survivorship bias.
– Timing and look-ahead bias: form portfolios using accounting data available at the formation date (often using last fiscal-year data available with appropriate publication lag).
– Weighting: choose value-weighted for market-representative factors, equal-weighted if you want size-neutral portfolios. Document choice.
– Frequency and formation rules: monthly rebalancing is common; some choose quarterly/yearly formation depending on accounting update frequency.
– Statistical diagnostics: check R-squared, t-statistics, residual serial correlation and heteroskedasticity (use robust standard errors).
– Robustness checks: test alternative sorting breaks, weighting, and sample periods. Consider out-of-sample or rolling-window regressions.
Interpreting results and practical uses
– Portfolio construction: use HML exposure to tilt a portfolio toward value or growth depending on investor preference and risk tolerance.
– Performance attribution: a positive contribution from HML in returns decomposition indicates exposure to the value premium rather than unique skill.
– Risk management: recognize that HML exposure can be cyclical — value underperforms during certain macro regimes (e.g., growth-favored markets).
– Expected return implications: historically, value stocks earned higher average returns in many markets, but premiums are neither guaranteed nor constant.
Limitations and caveats
– Time variation: HML premium varies over time and across countries — past performance does not guarantee future results.
– Measurement choices: how you measure book equity, rebalancing frequency, breakpoints, and weighting materially affect HML estimates.
– Transaction costs and capacity: implementing large value tilts can involve trading costs and liquidity constraints.
– Factor crowding: if many investors chase the same premium, returns may shrink and risk of large drawdowns may grow.
– Model specification: other factors (e.g., momentum, quality) also explain returns; Fama and French expanded the framework to five factors to capture profitability and investment effects.
FAQs
Why is Fama–French often considered better than CAPM?
– CAPM uses a single market factor and assumes market beta fully explains expected returns. Empirically, size and value effects (SMB and HML) improve cross-sectional explanatory power. Several empirical studies find the Fama–French three-factor (and subsequent five-factor) models explain average returns across portfolios better than CAPM. However, “better” depends on portfolio construction, sample, and which anomalies are included in the test set. (See Fama & French, 1993; related empirical studies.)
What does the HML beta mean?
– The HML beta measures sensitivity to the value premium. A positive HML beta means the portfolio behaves like a value portfolio (benefits when HML is positive). A negative HML beta means the portfolio is growth-tilted. The beta’s magnitude tells you the relative exposure; alpha measures residual abnormal return after accounting for HML and other factors.
Example interpretation
– If a portfolio regression yields βHML = 0.8 and α ≈ 0: the portfolio has a strong value tilt, and most of its excess return can be explained by exposure to the HML factor rather than manager skill. If α is significantly positive after controlling for HML and other factors, the manager may have genuine skill or the model may be missing relevant factors.
Selected sources and further reading
– Investopedia. “High Minus Low (HML).” (Source URL provided by user.)
– Fama, Eugene F., and Kenneth R. French. 1993. “Common risk factors in the returns on stocks and bonds.” Journal of Financial Economics, 33(1): 3–56.
– Fama, Eugene F., and Kenneth R. French. 2015. “A five-factor asset pricing model.” Journal of Financial Economics, 116(1): 1–22.
– International Journal of Business and Management. “CAPM vs Fama–French Three-Factor Model: An Evaluation of Effectiveness in Explaining Excess Return in Dhaka Stock Exchange.”
– Public and Municipal Finance. “The use of CAPM and Fama and French Three Factor Model: portfolios selection.”
– Provide code snippets (Python/pandas or R) to build HML and run the Fama–French regression on a sample dataset.
– Produce a compact checklist/template you can use when implementing factor construction in your shop.