Earnings Predictability of DuPont Factors: Impact of Mean Reversion and Competitiveness

We reexamine the predictive roles of DuPont decomposition components—asset turnover (ATO) and profit margin (PM)—in forecasting future operating profitability. We demonstrate that ΔPM is conditionally informative, rather than weak per se, by integrating mean reversion and competitiveness and connecting findings to earnings persistence and industry structure. The prior literature generally finds that changes in ATO dominate changes in PM toward predicting future return on net operating assets (ΔRNOA), despite theoretical arguments suggesting that both components should contain predictive information. Using a sample of 99,938 U.S. firm-year observations from 1981 to 2020, we reconcile these findings by incorporating heterogeneous mean reversion and firm competitiveness into the DuPont framework. We first document that PM exhibits substantially faster mean reversion than ATO, which weakens the unconditional predictive power of ΔPM. We then introduce industry-relative measures of ATO and PM to capture firms’ competitive positioning within industries. The results show that both ΔATO and ΔPM become significantly more informative for firms with stronger relative competitive advantages. In addition, firms with higher relative ATO and PM exhibit higher earnings and revenue persistence and are more likely to sustain consecutive earnings increases. We further find that competitiveness effects are stronger in diversified industries, while mean reversion is slower in concentrated industries. Overall, our findings provide a unified framework linking DuPont analysis, competitiveness, and profitability persistence, with important implications for forecasting, valuation, and financial statement analysis.