Measures of Firm Performance and Concentration: Stylized Facts and a Dilemma of Data Reproduction
Economists take it as given that the concentration of return on invested capital (ROIC) in the US increased over the last decades. Principle evidence to support this claim is a graph presented by the Council of Economic Advisors (CEA, 2016).
We indicate that the widely circulated graph cannot be reproduced based on Compustat data and established account measures and calculations. Further, we indicate that an arbitrary or ad-hoc selection of firms may have led to the initial results presented by the CEA. We criticize their lack of robustness and lack of communication of it as the story presented by the original authors is incomplete and potentially misleading.
Rather than rejecting the results of the CEA altogether, we correct their story in its core statement and contribute to it as well. As the CEA finds an increasing concentration in the upper end of the distribution, we find increasing concentration for low-ROIC firms. The concentration increase for the top 10\% of firms that we find is lower than the one presented by the CEA and limited to large firms. The presented results in this paper are satisfyingly robust, regardless of whether we use only `large firms’ or the complete, available sample of data.
We conclude that the original story of an increased concentration of ROIC must be handled with care. It is necessary to highlight the lack of robustness concerning the results presented by the CEA and amend the story with the increased concentration at the bottom of the distribution. We further make the case that all available data should be used in economic analyses rather than relying on arbitrary data selections. Latter may, as we show, lead to wrong or incomplete results, misleading the public and political debate about critical economic issues of our time.