Time to Unpack: Industry, Corporate, and Business Segment Effects on the Components of ROA

Abstract

Organizational scholars have long been interested in the influence of industry, corporate, and business segment effects on organizational performance. Research on this important topic has used return on assets (ROA) as the indicator of organizational performance, and when this research is considered in the aggregate, there appears to be a general consensus regarding the role of these multilevel influences. However, theories of firm boundaries and value creation suggest that multilevel effects may influence the two conceptually meaningful components of ROA—income and assets—quite differently. This, together with properties of ROA that complicate statistical analyses, may obscure our understanding of the drivers of organizational performance that exist at different levels. We first examined this issue with a simulation and found that when a stable between-level (e.g., firm) effect influences both the numerator (e.g., income) and denominator (e.g., assets) to the same degree, estimates of the between-level effect dissipate almost entirely when the two components are combined as a ratio (e.g., ROA). We then used Compustat Business Segment data to assess the generalizability of our findings to organizations. Consistent with the results of our simulation, stable firm- and business-level effects on income and assets substantially exceed the corresponding effects on ROA. Together, these results suggest that our theoretical understanding of organizational performance would be well served by research that considers income and assets separately.