sector is inversely related to that of secondary employment. For this reason, it could be argued that
the aforementioned results could be viewed only as partial evidence.
The theoretical part of the Holtz-Eakin and Lovely model, however, assumes that public
infrastructure would not have any significant impact on the non-manufacturing part of the economy.
In fact their results for the US economy have corroborated this thesis. Thus, the Greek results
presented in this section can be viewed as another confirmation of the model’s assumption, with the
caveat that the quasi-production functions used here cannot describe accurately this specific part of
the Greek economy.
4 Investigating Alternative Channels of Infrastructure Effects at Different Spatial Levels in
Greece
The direct impact of public capital on the manufacturing sector of the Greek economy has
been extensively analysed in Rovolis and Spence 1997a, 1997b, 1998, at national, regional, sectoral,
and urban levels. However, it seems worthwhile to seek for other potential channels by which
infrastructure can affect the private economy, perhaps in more subtle ways. The Holtz-Eakin and
Lovely (1996) model suggests that there is a variety of such channels at least from a theoretical
viewpoint. Their empirical work examined in depth two possible ways by which a change in public
capital stock can affect the secondary sector. The first is by altering the scale of production for each
manufacturing firm. The second is by influencing the equilibrium number of firms. Thus, their model
can help to assess if the public infrastructure affects private manufacturing, either by changing the
level or by altering the composition of its productive activity. These issues can be dealt with in turn.
4.1 Public capital’s effects on the preferred scale of production
The theoretical model described in section 2 postulates that the number of varieties of
intermediate goods in the economy can be used as a measure of the range of economic activity. The
greater the number of these varieties the more dynamic is the state of the economy (see equation 4).
The empirical counterpart of equation 4 is the following equation,
M a-1
(37)
— = n 1 x 0
n
which has been calibrated by Holtz-Eakin and Lovely using as the left hand side variable the output
per manufacturing establishment and the right hand side variables are the number of establishments
(as proxy of n) and public infrastructure (changes in infrastructure increase x0, see section 2) and with
the variables in logarithmic form. Thus, equation 37 becomes:
ln Г — ∣ = ln( na 1) + ln( x 0 )
< n J
(38)
18