R-Square and Market Efficiency


This paper addresses the debate about R-square as an indicator of information quality: Does low R-square indicate early resolution of uncertainty through the arrival of firm-specific information, or does it indicate a high level of uncertainty that remains unresolved? Tests based on the post-earnings-announcement drift, V/P, accruals, and net operating assets anomalies all reject the view that low R-square indicates a high quality information environment (early resolution of uncertainty). Low R-square firms have lower future earnings response coefficient, indicating that their current stock price incorporates a smaller amount of future earnings news, and thus more uncertainty about future earnings news remains unresolved. Furthermore, low R-square firms have worse information environment as measured by earnings quality, earnings persistence, and earnings predictability, and have higher probability of distress.