Competitive Strategy and Earning Persistence : Evidence from Large Sample based on Natural Language Processing and Deep Learning

As an important indicator of earnings quality, earning persistence is perceived as the reflection of operation management and the quality of internal control. It facilitates the optimization of resource allocation and provides investors valuable insights. Strategy is described as the long-term plan for firms and plays an influential role in the process of firm's internal control and financial performance. Competitive strategy is comprised of cost leadership strategy, differentiation strategy and focus strategy, where focus strategy is regarded as the application of cost leadership strategy and differentiation strategy in a focused market. Thus, in this paper, cost leadership strategy and differentiation strategy are mainly examined. We develop a textual measure of firm's competitive strategy based on natural language processing (NLP) and deep learning to expore the linkage between competitive strategy and earning persistence. We find that firms with differentiation strategy have higher earning persistence and are more sustainable. In addition, consistent with Porter's “Stuck in Middle” theory, firms with dual strategy are negatively associated with earning persistence. In summary, this paper contributes to the strategic management and earning persistence literatures by developing a new approach of strategy based on natural language processing and deep learning, which allows for conducting large-scale longitudinal analyses of its antecedents and consequences