Review of Industrial Organization 15: 149–163, 1999.
© 1999 Kluwer Academic Publishers. Printed in the Netherlands.
The Measurement of Firm Information About
Department of Economics, SUNY - Binghamton, Binghamton, NY 13902-6000, U.S.A.
Economics Department, University of Richmond, Richmond, VA 23173, U.S.A.
Abstract. The role of ﬁrm information about product and ﬁnancial markets is the subject of consider-
able research. Typically empirical research measures information through price dispersion. However,
the dispersion represents an imperfect measure of information. Several studies utilize stochastic
frontier estimation techniques to measure worker information about the labor market. This paper
determines whether the frontier information measure can be applied to the measurement of ﬁrm infor-
mation about product markets. Several intuitive hypotheses are tested concerning the relationship be-
tween ﬁrm characteristics and information investments. The results are consistent with expectations
and provide support for using stochastic frontier techniques to measure ﬁrm information.
Key words: Firm information, stochastic frontier estimation, uncertainty.
The effects of incomplete market information have received considerable atten-
tion since Stigler (1961). The majority of research concentrates on the theoretical
implications of incomplete information. Studies that consider this issue from an
empirical perspective include Stigler (1961, 1962), Mills and Schuman (1985), and
Ghosal (1991, 1995a,b). These studies use the standard deviation or a conditional
standard deviation as measures of incomplete information. For example, consumers
and ﬁrms have less information (or greater uncertainty) the greater price dispersion.
This paper discusses some problems associated with dispersion as an information
estimate and presents an alternative measure used to examine information in the
labor market. Our primary purpose is to determine whether the alternative measure
provides reasonable estimates of ﬁrms’ product market information.
While widely used, Gaynor and Polachek (1994) discuss two problems with the
dispersion measure. First, prices vary for numerous reasons and price dispersion
captures differences in many market factors in addition to information. Recent
studies partially address this issue. Ghosal (1995a,b) uses industry level data and
We wish to thank two anonymous referees and Geoffrey Shepherd for many helpful comments.