pon the intellectual traditions associated with asset rights and asset types In section 4, we describe how we use this typology to classify all the firms in COMPUSTAT between 1998 and 2002 based on the text of the SEC 10K filings. We report statistics to show that or classification has strong inter-rater reliability(some evidence of convergent validity) and is distinct from industry classification(discriminant validity). We also describe how we use ANOVA and variance decomposition methods-standard in the empirical literature--to analyze the extent to which business models matter in firm performance In section 5, we report our baseline results. The evidence is that business model effects are larger than year effects in explaining performance heterogeneity, as measured by return on assets(rOa)or return on sales(Ros). Importantly, business model effects also appear to be at least as strong, if not stronger, than industry effects in explaining performance In section 6, we report evidence that our interpretation is unlikely to be a result of reverse causality, or to systematic differences in the level of diversification in the firms in our sample. We also test the robustness of our finding to different interaction effects and treatment of outliers. And we address issues of potential sample selection bias and measurement errors. We find that our baseline conclusion This study does not claim that our definition of business model is unique, although we argue that it satisfies important criteria. Like the empirical literature on firm-versus-industry effects, we also have not answered questions like how business models impact performance, nor do we address the normative question of how individual firms can exploit or modify their business models to improve their performance. We hope that the work described here provides a foundation for future work on these questions 2. MOTIVATION AND ANTECEDENTS The"industry view"of performance heterogeneity among firms is usually associated with industrial organization(10).(Porter, 1980) develops the early IO structure-conduct-performance framework into a2 upon the intellectual traditions associated with asset rights and asset types. In section 4, we describe how we use this typology to classify all the firms in COMPUSTAT between 1998 and 2002 based on the text of the SEC 10K filings. We report statistics to show that our classification has strong inter-rater reliability (some evidence of convergent validity) and is distinct from industry classification (discriminant validity). We also describe how we use ANOVA and variance decomposition methods—standard in the empirical literature—to analyze the extent to which business models matter in firm performance In section 5, we report our baseline results. The evidence is that business model effects are larger than year effects in explaining performance heterogeneity, as measured by return on assets (ROA) or return on sales (ROS). Importantly, business model effects also appear to be at least as strong, if not stronger, than industry effects in explaining performance. In section 6, we report evidence that our interpretation is unlikely to be a result of reverse causality, or to systematic differences in the level of diversification in the firms in our sample. We also test the robustness of our finding to different interaction effects and treatment of outliers. And we address issues of potential sample selection bias and measurement errors. We find that our baseline conclusion is robust. This study does not claim that our definition of business model is unique, although we argue that it satisfies important criteria. Like the empirical literature on firm-versus-industry effects, we also have not answered questions like how business models impact performance, nor do we address the normative question of how individual firms can exploit or modify their business models to improve their performance. We hope that the work described here provides a foundation for future work on these questions. 2. MOTIVATION AND ANTECEDENTS The “industry view” of performance heterogeneity among firms is usually associated with industrial organization (IO). (Porter, 1980) develops the early IO structure-conduct-performance framework into a