Current Agriculture, Food & Resource Issues
C. E. Ward
Summary of the Evidence and Concluding Comments
Tables 1 to 4 summarize the research reviewed in this paper. One point is clear. Research
varies widely in terms of data, i.e., data unit aggregation (transactions to annual
observations), collection length (one month to decades), and spatial aggregation (local market
to the entire United States), as well as methodological approach, i.e., econometric estimation
of models with several functional forms, simulation, game theory, conjectural variation, and
combinations thereof.
Azzam and Anderson (1996) conducted an extensive review of competition in
meatpacking. In their summary, they offered criticisms both of the structure-conduct-
performance approach and the conjectural variation approach. They concluded the body of
empirical evidence was insufficient to argue persuasively the meatpacking industry was not
competitive. Sexton (2000) reported on more recent critiques of the conjectural variation
approach. Despite the weaknesses of the approach, he concluded that market power estimates
in meatpacking (i.e., the focus of much of the conjectural variation literature) are modest,
and structural changes on balance are probably beneficial from an efficiency viewpoint.
In reviewing the available evidence, my interpretation appears consistent with theirs.
Examining the evidence in the tables, either by data aggregation, methodology, or time
period, yields little difference in interpretation. Research shows a dynamic, bi-directional
linkage between structure, conduct, and performance. This exhibits itself in the research that
measures indirectly the linkage between structure and performance as well as in the research
that measures directly the linkage between behavior and performance.
It seems both the structure-conduct-performance approach and the conjectural variation
approach share a related weakness. In the Bainsian structure-conduct-performance approach,
excessive emphasis is placed on a single structural characteristic, i.e., the concentration ratio
or HHI, as a predictor of conduct and performance. In the conjectural variation approach,
emphasis shifts to a single conduct variable, i.e., the conjectural variation coefficient, as a
descriptor of conduct and predictor of performance. In either case, relatively little emphasis is
placed on the “how” and competitive dynamics, especially of rivals’ reactions.
Besides the “how” of exercising market power, two additional central issues emerge. The
first centres on the fact that most of the research leads to questions of the form, “How large is
large?” or “How small is small?” Price distortions of 3 percent or less were found in most
studies. These fall well short of regulatory agency standards related to merger impacts and
non-competitive behavior, which often assume a 5 percent price impact rule (U.S.
Department of Justice and Federal Trade Commission, 1997). However, the courts and
regulatory agencies have defined specifically neither how much market power is “significant”
nor for how long a firm or firms must maintain significant market power (Carlton and
Perloff, 1994).
From another viewpoint, seemingly small impacts on a $/cwt. basis may make a
substantial difference to livestock producers and rival meatpacking firms operating at the
19