Review of Industrial Organization 15: 89–92, 1999.
Technology and Market Structure. Theory and History. John Sutton. Cambridge,
Massachusetts: The MIT Press, 1998, 676 pages, $55.
Game theory has strongly inﬂuenced the theory of industrial organisation over the
last twenty years. Despite many useful new insights, game theory has been less suc-
cessful in providing veriﬁable hypotheses for empirical investigation with a general
validity across a broad range of industries. The models come with equilibrium
outcomes that show little robustness to slight changes in the basic assumptions.
Thus, the empirical testing of game theoretic models is typically conﬁned to single
industries with limited scope for more general validity.
This book addresses these issues by suggesting a broad theoretical framework
that is compatible with the equilibrium outcomes of a class of game theoretic mod-
els. The study also sets out to reconcile two lines of explaining market structure
that have been kept separate in the literature: on one side there is the cross-section
approach in explaining market structure, as indicated by Bain’s classical structure-
conduct-performance paradigm; on the other side is the intertemporal dimension
of the growth of ﬁrms literature, based on Gibrat’s law of random growth of ﬁrms.
Overall, the task is daunting, but the author provides convincing answers to the
questions he poses. By developing the ‘bounds approach’, he shows how much
mileage one can get from simple concepts borrowed from game theory in setting
up a broad theoretical framework able to deliver empirically veriﬁable hypotheses.
The study concentrates on R&D intensive industries. The theory is matched with
empirical evidence from cross-section analysis and by using individual industries
as ‘natural experiments’.
The bounds approach consists of a set of assumptions that encompass a class of
models which can be tested by putting observable constraints on the data. This is
different from the traditional approach based on a single, complete model identify-
ing some unique equilibrium outcome. With the traditional approach, the empirical
IO economist is actually imitating the economic historian by trying to identify facts
that explain a particular event but are unable to formulate propositions that hold for
a broad range of industries. The bounds approach therefore is much wider in the
model deﬁnition, but it nevertheless restricts the outcomes in a way that is tight
enough to be interesting. The space of outcomes is partitioned into two sets: those