a whole, and for the Greece excluding the Athens area, there is a statistically significant impact of high
magnitude.
Following Holtz-Eakin and Lovely (1996), the above results were double-checked by a second
set of regressions, in which it was assumed that public capital should influence the total manufacturing
output via its impact on output per firm and the number of varieties. The results for these latter
regressions corroborate those obtained earlier.
The second channel by which infrastructural change affects private manufacturing, in the
context of this model is the concomitant changes in the equilibrium number of manufacturing
establishments. The direct regression of the number of manufacturing establishments on the public
capital variable (either in its total, productive, or social form) has shown that there is no such direct
impact, whatever the spatial level. The next step was to augment these regressions with the private
inputs of production. Once again the results obtained generally showed that there was little significant
impact of the infrastructure variables.
One conclusion that can be drawn from comparing these results for Greece with the findings
from the US research is that for the more developed economy infrastructure works more via its effects
on the composition of the manufacturing activity, whereas in the Greek case it seems to affect more
directly the level of this activity. However, it has to be kept in mind that the US case refers to a huge
economy in comparison to the Greek one, and that the US empirical work is based on four cross-
sectional surveys articulated into a panel, while the Greek panel has a significantly longer time
dimension. In any case, both of these empirical works show that public capital seems to have little if
any impact on the non-manufacturing part of the economy. However, the results for the Greek case
must be viewed with the caveat that crucial data for this part of the analysis were unavailable, and that
it was conducted in a rather indirect way (quasi-production functions).
Finally, it must be noted that the whole analysis was conducted within the analytical
framework set by the Holtz-Eakin and Lovely (1996) model. This means that the results are as good
as the model’s assumptions, and must be viewed in this spirit. For instance, the results for the impact
of public capital on the equilibrium number of firms must not be construed as an attempt to evaluate
the infrastructural role on new firm creation. There is now an existing body of work on this topic for
Greece, and some of this research has incorporated the infrastructure variable into the analysis20. The
results of this empirical work have shown that public capital, does indeed play a positive role in new
firm creation, with a significant time lag (see Fotopoulos 1998). This implies that the model used here
has not allowed for such lagged impact of infrastructure on the equilibrium number of firms.
20 For a summary of the existing bibliography on the topic, see Fotopoulos (1998); for a different perspective on the
subject, see also Katseli (1990).
29