Public infrastructure capital, scale economies and returns to variety

where Q is total manufacturing output (GPV), K/n is private capital per firm, L/n is labour per firm, n
is the number of firms, and G is public infrastructure capital.

Even though in the case of Greece infrastructure capital seems to play a significant role in the
determination of the output per manufacturing establishment, it is still useful to extend the analysis in
this context as per Holtz-Eakin and Lovely. The results would then corroborate or refute the findings
of the previous section. Equation 40 changes for the Greek model to:

Q_it ~ —ⁱ~ +—ⁱ~ + nit + Git +1 + uit

it                                 it it               it

it

(notation as in previous equations)

Again there are four different levels of spatial analysis. The estimations for the regional panel are
given in table 8. The coefficient for total public capital, in the first regression of the table, is positive
and statistically significant. Its magnitude is high (0.254), as is also the case for the coefficient of
labour. However, the coefficient for private capital is small (0.046) and statistically significant only at
ten percent level. The degree of homogeneity a (see section 2) is the magnitude of the estimated
coefficient of the manufacturing establishments (this coefficient corresponds directly to a, in contrast
to the previous section where the estimated coefficient was a-1). Here, a is equal to 0.761
(statistically significant) and is larger than the respective figure obtained in the previous set of
regressions. The next two regressions present the results for the breakdown of public capital into
productive and social categories respectively. It is interesting that at regional level productive
infrastructure seems to account for practically all of the impact of total public capital. The results for
this type of infrastructure are almost identical to those for the total. However, it seems that the social
component has practically zero impact, as its coefficient is small and statistically insignificant (-0.048).

Table 8 Infrastructure effects on total output: Regional panel for total manufacturing, 1982-1991

Dependent Variable: ln of Total Manufacturing Output (GPV)

*** Statistically significant at 1% level, ** Statistically significant at 5% level, * Statistically significant at 10% level

The results for the sectoral panel for the whole of Greece are not essentially much different
(table 9). In the regression incorporating total infrastructure the coefficient for private capital appears
to be statistically insignificant, that for labour is more or less the same magnitude (0.722 and
statistically significant) as for the regional panel, and similarly for the manufacturing establishment
variable (0.785 and statistically significant). The estimate for infrastructure is, nevertheless, much

22

<<   <   19   20   21   22   23   24   25   >   >>