Concepts‐based Accounting Standards
用机器学习中的S形曲线分析基于概念的会计准则为何无法保证可比性，且其带来的一致性未必总是可取的，为会计标准分析和实证研究提供新框架。
While comparability across firms and consistency over time are generally held to be fundamental goals of financial reporting, I provide an analytic representation of a concept that explains why concepts‐based accounting standards cannot assure comparability and why their induced consistency may not always be desirable. While the term ‘concepts‐based accounting standards’ has not caught on in the academic and professional literatures, its use here emphasizes the foundational role that language‐based concepts play in constructing accounting standards. I appeal to the academic literature in machine learning, neural networks, and especially cognitive science—all of which may represent concepts by ‐curve (sigmoid) signatures. I then show how ‐curves can explain an accounting standard's (1) precision, (2) comparability across firms, (3) demands placed on judgement, and (4) consistency across time. Accordingly, an ‐curve formulation may guide both analytical modelling of accounting standards and add structure to empirical research designs.