Topic 1: How Does Industry Matter?
1. Complete Reference of the Article:
Rumelt, R. P. 1991. How much does industry matter? Strategic Management Journal, 12: 167-185.
2. Brief Synopsis of the basics of the paper:
2.a. Theoretical Argument:
Opposed to one of the most relevant assumptions of industrial organization economics (The neoclassical
model of industry) in which the dominant unit of analysis has been focused on the industry level, Rumelt
approaches the field of business strategy posing that differences among firms arise from the specific and
unique circumstances and actions of individual firms or business-units.
2.b.Assumptions:
The business strategy approach seems to explain better the importance of industry than industrial
economics.
There are stable and fluctuating effects that influence the industry effects on business-unit returns.
2.c. Major Proposition:
When stable and fluctuating effects are considered in order to explain the industry effects on business-unit
returns, the conclusions of traditional analyses change radically.
2.d.Research Methods:
The data used comes from FTC’s Line of Business Program that gathers operations of large U.S.
corporations with disaggregate data on the profits of corporations by industry. Given that the heterogeneity
within industries is crucial in this paper, Rumelt performs a variance components model, more specifically,
Searle’s treatment of the theory and practice of variance component estimation method.
2.e. Main Empirical Results:
Practically all the observed differences among industry returns are unrelated with long-term industry effects
but to the random distribution of high and low-performing firms across industries.
2.f. Theoretical Implications:
Given that industries are heterogeneous, the focus on industry analysis must be changed. Furthermore, long-
term rates of return are not associated with industry but with exclusive situations, strategies and
performances of individual firms.
2.g.Empirical Implications:
Firstly, stable industry effects account for a very small percent of the variance in business-unit returns. The
same situation holds with the dispersion in industry returns but with a slightly higher proportion. Secondly,
most of the ‘residual’ variance is based on the stable component and the long-term differences among
business-units instead of the fluctuating component. However, perhaps the most important empirical
implication is that business-units within industries differ from one another much more than industry among
themselves. Some secondary implications are:
1. Models that include industry as unit of analysis can at best explain only 8% of the dispersion among
business-unit profit rates. Similarly, if we use corporation as unit of analysis, it can at best explain
2% of dispersion.
2. When trying to explain this dispersion the best unit of analysis to be analyzed is the business-unit or
smaller entities focusing on the source of heterogeneity within industries rather than relative size.
3. Assessment of strengths/weaknesses of the paper:
Strengths:
It improves the variance components analysis performed by Schemalensee in 1985 by analyzing four
years instead of one. Additionally, the sizes of both samples analyzed are also large enough to draw
some consistent empirical conclusions. Finally, this variance model includes year-to-year variations
in overall returns and year-to-year variations in industry-specific returns.
Weaknesses:
Some of the FTC 4-digit industries may be too broad to reveal true strength on industry effects. Also,
given that it represents one of the pioneers that changed the unit of analysis from the industry-level
to the business-unit, results are only partially exportable to different contexts.
Significant BU effects, much higher than industry and
corporate effects
