The profitability premium is the tendency for firms with higher profitability to earn higher average stock returns, despite having higher valuations.
Measurement
Gross profitability (GP/A): revenue minus cost of goods sold, scaled by total assets. Novy-Marx (2013) argues this is the cleanest measure because items further down the income statement are polluted by expensed investments (R&D, advertising, human capital development) that reduce reported earnings without reflecting true economic productivity.
Operating profitability (OP): used in the Fama-French five-factor model to construct RMW (Robust Minus Weak). Measured as revenue minus COGS, minus SG&A, minus interest expense, scaled by book equity.
QMJ profitability: Asness, Frazzini, and Pedersen (2019) use a composite of gross profits over assets, ROE, ROA, cash flow over assets, gross margin, and low accruals.
Key empirical evidence
- Long-short GP/A strategy earns 0.31%/month raw and 0.52%/month FF3 alpha (t=4.49)
- Comparable in power to book-to-market (0.41%/month)
- Significant even among large caps (0.26%/month) and internationally (0.76%/month across 18 developed markets)
Theoretical rationale
From the dividend discount model: holding book value and growth constant, higher expected earnings imply a higher discount rate (required return). Productive firms that investors demand high returns to hold are priced similarly to less productive firms requiring lower returns. Variation in productivity thus identifies variation in required rates of return.
This argument is consistent with both rational pricing and behavioral mispricing.
Relationship with value
Profitability and value are negatively correlated (-0.57). Profitable firms tend to be growth stocks (low BE/ME, high valuations). This makes profitability strategies a natural hedge for value:
- Combining them yields a Sharpe ratio of 0.85 (vs. 0.34 for the market)
- In double sorts, controlling for profitability raises the value spread from 0.41% to 0.68%/month
- Controlling for BE/ME raises the profitability spread from 0.31% to 0.54%/month
Novy-Marx characterizes profitability as “the other side of value”: while value strategies acquire inexpensive assets by selling expensive ones, profitability strategies acquire productive assets by selling unproductive ones.
Challenges to distress-based explanations
The profitability premium undermines popular explanations of the value premium:
- Profitable firms are less prone to distress, yet earn higher returns
- Profitable firms have longer cash flow durations (contradicting duration-based explanations)
- Profitable firms have lower operating leverage (contradicting operating leverage models)
Cash flow measures as predictors
Foerster, Tsagarelis, and Wang (2017) show that direct cash flow measures predict stock returns better than income statement profits. They transform indirect method cash flow statements into a standardized direct method template that disaggregates cash flows by economic characteristics (operating, financing, tax, and extraordinary activities).
Key findings:
- Stocks in the highest direct-method cash flow decile outperform the lowest by over 10% annually on a risk-adjusted basis
- Direct cash flow measures beat indirect cash flow measures, which in turn beat income statement profitability measures (gross profits, operating profits, net income)
- Cash taxes and capital expenditures provide incremental predictive power beyond operating cash flows
- Results are robust across investment horizons, risk factor models, and sector controls
This extends Novy-Marx’s (2013) insight that items further down the income statement are “polluted” by suggesting that cash-based measures sidestep accrual manipulation entirely. The correlation between direct method operating cash flow and standard free cash flow to equity (FCFE) is only 45% when scaled by total assets, indicating these capture substantially different information.
Four-factor model
Novy-Marx (2013) proposes a four-factor model using market plus industry-adjusted value, momentum, and profitability (PMU) factors. This model prices 15 anomaly strategies with RMSE of only 0.22%/month, less than half the Fama-French four-factor model’s 0.54%.