Industry and Firm Effects

Contributed by: Anita McGahan

Since the publication of Richard Schmalensee’s 1985 paper on industry and corporate effects on performance, scholars in the field of Strategic Management have written and published a series of papers examining how firm performance – typically represented as return on investment – relates to the firm’s participation in particular industries as compared to the idiosyncratic influence ­of the firm itself distinct from competitors. By describing the relative importance of industry and firm effects, the literature seeks to identify whether managerial intervention through strategic action has a significant impact on firm performance compared to the effects of industry structure, which are often construed as mainly outside the control of corporate management.

There are several characteristics of this literature that are notable. The major studies in this line are descriptive rather than tests of well-defined theories, although a vigorous debate has emerged (see Hansen & Wernerfelt (1989), Roquebert, Phillips & Westfall (1996), McGahan and Porter (2002, 2005), Ruefli and Wiggins (2003, 2005)) about whether inferences are appropriate about causality. For an interpretation of the meaning of the effects based on their qualitative importance, see McGahan (1999b).

The principal empirical strategy is to examine a panel of business units or corporations using simple statistics on variance. The earliest papers in the line (see Rumelt 1991), McGahan & Porter (1997), Bowman & Helfat (2001), and Brush & Bromiley (1997)) evaluated whether random-effects or fixed-effects models were appropriate for identifying industry and firm effects. The more recent literature in the line acknowledges that the effects are nested and models the relationships between the effects directly (Hough (2006), Brush Bromiley & Hendrickz (1999), McGahan & Porter (2003)). A new approach uses rank ordering rather than variance to evaluate the influence of industry and firm effects on performance (Ruefli & Wiggins, 2003).

Some of the studies decompose variance at the level of the business unit (Rumelt (1991), McGahan and Porter (1997, 1999, 2002, 2003) while others decompose variance at the level of the corporation as a whole (McGahan (1999a), Khanna & Rivkin (2001), Bowman & Helfat (2001)). When performance is modelled for the corporation as a whole, the corporate effects must be interpreted differently than when business-unit effects are also included (McGahan & Porter (2002), Bowman & Helfat (2001), McGahan (1999b).

A parallel line has emerged on how the effects of industry and of the firm change over time (Mueller (1986), Ghemawat (1991), Waring (1996), McGahan & Porter (1999, 2003), Furman & McGahan (2001), Ruefli & Wiggins (2003)). While the precise empirical design of these studies differs significantly, the main idea in them is to identify which of the effects on performance tend to be most stable over time and to make inferences about their relative importance based on their stability.

Many of the studies in this literature rely on the Compustat business-segment (or “industry-segment”) reports, which relied on an accounting requirement in place from about 1980 to 1997 that firms report by line of business. Many of the companies covered in the reports changed their accounting conventions after about 1997 (and especially after 1999) when this requirement was liberalized. Also note that, in 1993, a change in pension accounting requirements led to the creation of many anomalous, temporary business units. Several studies have sought to overcome the disadvantages of the Compustat reports by applying decomposition techniques to other measures of performance (Chang & Singh (2000), McGahan (1999a)).

The main findings in the literature are that industry, corporate, and business-unit are all important to performance, although there is disagreement about which effects are most significant and why. The disagreements rest on questions about noisiness in the data that obscures industry effects, methods, etc. (McGahan & Porter (2005), Ruefli & Wiggins (2005), Hough (2006)). There is also evidence that all three types of effects are remarkably persistent, and that idiosyncrasies in performance are generally persistent. These results provide strong support for the idea that differences tend to be sustained regardless of their origins in industry and firm factors.

New studies in this line tend to fall in one of three categories. First, some scholars seek to establish the relative importance of geographic differences to performance by decomposing the variance of the performance of firms located in different parts of the world. The leading research in this area distinguishes different kinds of country effects, eg, the host-country effect (Makino et al (2004)) vs. the home-country effect (McGahan and Victer (2009)). Second, a line of research investigates in detailed case studies how industry and firm effects co-evolve, especially in the early phases of industry development or during periods of industry disruption. Leading papers consider how individual entrepreneurs shape institutions that create enduring industry effects (Kaplan and Murray (2009)). Finally, a final set of studies is investigating the nature of competition and the consequences for how industry and firm influences on performance interact over both the short and long term. The leading research in this line shows how early competitive interaction sets behavioural precedents that are enduring (Klepper (2008), Feldman and Lowe (2008)).