Review of Industrial Organization 17: 441–456, 2000.
© 2000 Kluwer Academic Publishers. Printed in the Netherlands.
Why Do Established Firms Enter Some Industries
and Exit Others? Empirical Evidence on Italian
ALESSANDRO SEMBENELLI and DAVIDE VANNONI
University of Turin and CERIS-CNR
Abstract. In this paper the determinants of entry and exit decisions are analysed empirically on a
sample of well established business groups operating in Italian manufacturing. The focus is on the
role of sunk costs as entry barriers. Two competing hypotheses are tested. On the one hand, setup
costs, R&D and advertising outlays act as barriers to entry for both new and already established ﬁrms
because of their industry-speciﬁc commitment value. On the other hand, they may induce established
ﬁrms, which operate in similar industries, to enter. This is the case if R&D and advertising are ﬁrm-
speciﬁc investments which generate externalities to be efﬁciently exploited in adjacent industries.
Overall results suggest that the second hypothesis gives a better picture of the behaviour of our
sample of ﬁrms.
Key words: Entries, exits, sunk costs.
Jel Classiﬁcation: L10, L20.
The objectives of this paper are twofold. First, we provide novel descriptive evid-
ence on entries and exits of the leading ﬁrms in Italian manufacturing. Second, we
present estimates of two empirical models where entry and exit decisions are func-
tions of the characteristics of four types of factors: ﬁrm-speciﬁc variables, origin
and target industry characteristics, and measures of similarity between origin and
This work clearly ﬁnds its natural place among the studies on entries and exits
but, differently from the “main stream view”, the analysis is at the ﬁrm level and
does not focus on differences in entry/exit rates across industries. Also, we consider
a sample of large diversiﬁed ﬁrms, while traditional studies concentrate on small
The authors would like to thank S. Davies, P. Geroski, two anonymous referees, and the par-
ticipants to the EARIE Conference in Vienna where a preliminary draft of this paper has been
presented, for useful comments and suggestions.
Seminal studies include, among others, Dunne et al. (1988), Siegfried and Evans (1994), Wagner
(1994) and the papers collected in Geroski and Schwalbach (1991).