Review of Industrial Organization 18: 183–200, 2001.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands.
Firms’ Cost Functions: A Reconstruction
RICHARD A. MILLER
Department of Economics, Wesleyan University, Middletown, Connecticut 06459, U.S.A.
Abstract. Three empirical stylized facts characterize a ﬁrm’s cost functions: AVC is generally hori-
zontal; scale economies are often important; and capital usage (as well as labor usage) varies with the
rate of output. In contrast our standard theory of diminishing marginal products, variable proportions,
and U-shaped unit cost curves does not adequately reﬂect these stylized facts of industrial reality.
Explicit distinction between capital stocks and capital ﬂows allows a remodeling of a ﬁrm’s cost
curves to conform with these empirical facts.
keywords: Capital ﬂows, cost curves, marginal productivity theory, production functions.
JEL Citation Index: A22, D24, L2.
What do we know generally about production functions and cost functions for
individual ﬁrms? The following stylized facts appear to be valid generalizations
of a wide range of empirical work:
First, average variable cost functions are generally horizontal up to capacity.
Virtually every empirical study of existing plants produces this result (Bain, 1948;
Dean, 1976; NBER, 1943; Walters, 1963; Hall, 1986; Johnston, 1960; Kahn, 1989
(1929); Mansﬁeld, 1988). Moreover, a recent survey of executives’ beliefs about
their short run variable costs indicates that “only 11 percent of GDP is produced
under conditions of rising marginal cost” (Blinder et al., 1998, p. 102). Of the
remainder 48% is produced under constant MC and over 40% under falling MC.
Second, some and perhaps many long run cost (and production) functions ex-
hibit economies of scale. Empirical studies reveal signiﬁcant economies of large
scale in for example pipelines (Cookenboo, 1955), automobiles (White, 1971), and
a variety of other industries (Scherer et al., 1975; Pratten, 1971).
Third, for any given plant, factor usage differs as (the rate of) output is altered.
To reduce output, labor is laid off. So too is the plant. The ﬂows of all factor ser-
vices ( not just labor services and raw materials) change when output is increased
or reduced. Several studies conclude that this is the case in speciﬁc industries,
I am grateful for the helpful comments of and discussions with Gary Yohe, Gil Skillman, and
several referees, none of whom bears any responsibility for the remaining shortcomings, and for the
help of Gloria Cone and Vanita Gaonkar.