the same industry, cause the industry to concentrate in one single region even though the
individual firms have different individual preferences. The reason is that once one region has
attracted slightly more entrants than other regions, a critical threshold is passed, and suddenly all
firms will opt for this one region: a case of spatial lock-in.
In an empirical context, the outcomes of the spin-off model are not easy distinguishable from
the outcomes of the agglomeration economies model. We have, indeed, two different
explanations for the same phenomenon of spatial concentration of an industry. As spin-off
dynamics and agglomeration economies may well contribute to spatial concentration
simultaneously, the challenge for empirical research is to disentangle both processes so as to
assess their presence and importance. One out of the few studies that have attempted to do so is
Klepper’s (2001) study of the U.S. automobile industry. In his econometric analysis, he included
a dummy for being located in the Detroit area. The dummy showed no positive effect on the
survival of firms, which suggests that agglomeration economies were not present. The use of a
Detroit control variable, however, can be questioned, since a subset of firms within the Detroit
area may have benefited from each other’s presence through local networks (Giuliani,
forthcoming) or firms may have benefited from knowledge spillovers over a longer distance
(Jacob and Los, forthcoming). Despite this shortcoming, the result by Klepper (2001) strongly
suggests that the concentration of the U.S. automobile industry in Detroit can be attributed
mainly to the self-reinforcing dynamics of successful firms creating successful spin-offs, creating
successful spin-offs, et cetera.
A study by Boschma and Wenting (2005) on the spatial evolution of the British automobile
sector came to similar conclusions regarding the self-reinforcing nature of spin-off dynamics,
which, in the British case, led to a concentration in the Birmingham-Coventry area. However,
Boschma and Wenting also accounted for the presence of related industries (such as coach and
cycle making) in a region as a potential source of agglomeration economies, which was shown to
have a positive effect on the survival rate of firms. Thus, the local presence of related industries
appeared to be beneficial due to, for example, knowledge spillovers and skilled labour, yet the
local presence of a high number of firms operating in the same industry turned out to be harmful
due to increased competition, lowering the survival chances of new entrants. Another recent
elaboration on Klepper’s model is by Cantner et al. (2005), whose methodology using
instrumental variable estimation allows for post-entry innovation. In doing so, the survival