- 1Hou, Xue, and Zhang (2015) proposed a four-factor model — market, size, investment, and profitability — motivated by the neoclassical q-theory of investment, not by empirical fishing.
- 2Their "investment" factor (low-investment minus high-investment firms) and "ROE" factor (high-profitability minus low-profitability firms) together digest most of the standard anomaly zoo, including momentum-adjacent effects that the Fama–French three-factor model could not price.
- 3Where the Carhart four-factor or Fama–French three-factor model leaves large unexplained alphas on anomaly long-short portfolios, the q-factor model often reduces those alphas to economically small and statistically insignificant residuals.
- 4BCR uses investment discipline as a 10% weight inside its composite score — a direct operational nod to HXZ, who argue that firms expanding assets aggressively tend to underperform conservative capital allocators.
#The Paper at a Glance
Title: Digesting Anomalies: An Investment Approach Authors: Kewei Hou (Ohio State), Chen Xue (Cincinnati), Lu Zhang (Ohio State) Published: Review of Financial Studies, 2015
The paper's ambition is unusual for empirical asset pricing: instead of reporting a new anomaly and arguing for its uniqueness, HXZ set out to digest the existing catalog. They start from q-theory — the neoclassical proposition that a firm's optimal investment equates marginal q to marginal cost — and derive two first-order predictions. First, holding profitability fixed, firms that invest more should earn lower expected returns. Second, holding investment fixed, more profitable firms should earn higher expected returns. The authors then build a parsimonious four-factor model around those two predictions and show it prices most of the well-known anomaly portfolios.
#What the Paper Found
The Model Construction
HXZ construct factors from 2x3x3 sorts on size, investment-to-assets (I/A), and return on equity (ROE). The investment factor, often denoted r_IA, is the average return on the three low-investment portfolios minus the three high-investment portfolios. The profitability factor, r_ROE, is constructed symmetrically using ROE terciles. The market factor and a size factor round out the four-factor structure.
Anomalies Digested
The paper catalogs a large set of cross-sectional anomalies — momentum, earnings surprise, financial distress, accruals, net stock issues, asset growth, gross profitability, and many more — and asks which model best prices the long-short extremes. The qualitative pattern is consistent: the Fama–French three-factor model leaves economically and statistically significant alphas on most anomaly portfolios, while the q-factor model reduces most of those alphas to small, insignificant residuals.
| Anomaly category | FF3 alpha (long-short) | Q-factor alpha (long-short) |
|---|---|---|
| Momentum family | Large and significant | Reduced, often insignificant |
| Investment/asset growth | Large and significant | Near zero by construction |
| Profitability family | Significant | Near zero by construction |
| Financial distress | Significant | Substantially reduced |
The exact magnitudes differ across subsamples and anomaly definitions; the point is that q-theory's two economic levers do a lot of empirical work.
Economic Interpretation
The paper's framing is deliberately structural. The investment factor is not a "low-capex factor" in the descriptive sense — it is a proxy for differences in firms' implied discount rates. High-investment firms reveal that they find many projects attractive at their cost of capital, which (under the optimality condition) implies a lower cost of capital. The ROE factor plays the symmetric role on the profitability side.
#The Math (Lite)
The core expected-return equation, written in the paper's notation, is:
E[r_i - r_f] = β_MKT · E[MKT] + β_ME · E[SMB_q] + β_IA · E[r_IA] + β_ROE · E[r_ROE]Where: - r_IA is the investment factor, long low-investment and short high-investment firms. - r_ROE is the profitability factor, long high-ROE and short low-ROE firms. - The intuition from q-theory: optimal investment implies that discount rate ∝ profitability / investment. A firm investing heavily relative to its profitability has a lower discount rate and thus a lower expected return; a highly profitable firm that invests conservatively has a higher one.
#How Blank Capital Research Uses This
HXZ is the cleanest theoretical basis for two of BCR's composite factor sleeves.
- Investment (10% weight): We penalize firms that are growing their asset base aggressively without commensurate profitability. This is a direct operationalization of the q-factor intuition that high I/A firms carry a lower implied discount rate.
- Quality / profitability (30% weight): The ROE factor overlaps heavily with our quality sleeve, which uses gross profitability, return on invested capital, and accruals-adjusted earnings. HXZ motivates why profitability is a priced characteristic and not just a historical artifact.
Together these two sleeves carry 40% of the composite — the single largest bloc — reflecting the empirical weight HXZ's evidence places on the investment-and-profitability pair.
#Practitioner Watch-Outs
- Definition sensitivity. Investment-to-assets is not a primitive; there are several reasonable ways to compute it (change in gross property plant and equipment plus change in inventory, scaled by lagged assets, is one common choice). Small definition changes can matter for small-cap portfolios.
- Microcap contamination. As with most US factor research, a meaningful fraction of the raw spread lives in names that are effectively untradeable at scale. Robust implementations winsorize and weight by market cap.
- Profitability measurement. ROE is balance-sheet noisy. Many practitioners use gross profitability (revenue minus cost of goods sold, scaled by assets) following Novy-Marx for a cleaner signal.
- Factor correlations. The q-factors are correlated with Fama–French's RMW and CMA by construction. Running horse races without understanding this overlap can produce misleading conclusions.
- International evidence. The base paper is US-centric. International replications are broadly supportive but the investment factor is weaker in some regions with different capital-allocation norms.
#See It in Action
#Further Reading
- Hou, Mo, Xue, Zhang (2021) — a follow-up that augments the original four-factor model with an expected-growth factor, tightening the fit against forward-looking anomalies.
- Fama, French (2015) — the five-factor model published in the Journal of Financial Economics, which independently added profitability and investment factors and arrived at similar conclusions.
- Novy-Marx (2013) — the gross profitability paper that set the stage for profitability as a standalone priced factor.
#Related Factor Explainers
More on the quality literature.
Novy-Marx, Ball et al., Asness-Frazzini-Pedersen QMJ. The Learn library covers them all, and the Capital Memo threads them through the week as papers and markets collide.
Last updated · April 21, 2026