belonging to a group and a value of zero for those not belonging to any group, which, in previous
work (see for instance Parisi, Schiantarelli and Sembenelli, 2006), has been shown to be a
statistically significant correlate of firms’ productivity performance.
The innovation dummy reports a coefficient of 15.12 with 5% statistical significance. This means
that there exists a significant drift of 15.1 percentage points for productivity growth for the Italian
innovative firms, even in the years considered. This is particularly true for those firms introducing
process innovation (17.2%), while the drift is nearly not significant for those firms introducing
product innovation.
For the group of innovative firms, irrespectively of how innovation is defined, the estimate of the
age coefficient is negative and statistically significant. The point-wise estimates indicate that an
increase in the average age of the board members of one year translates into lower productivity
growth of some 0.28-0.37 percentage points. As to the non innovative group, the effect is zero:
having a younger board of governance on average does not spur productivity growth for the non-
innovative firms. The semi-elasticity of age for the whole sample is instead not significantly
different from zero. In order to capture the partial correlation of age and productivity growth it is
thus crucial to distinguish between innovative and non-innovative categories of firms.
The coefficient estimate for the temporary workers share is instead statistically significant and
negative both for the innovative and the non-innovative group, with point-wise estimates around -
.10 for the innovative and around -.20 for the non innovative firms. The Wald test for equality of
parameters in the case of the temporary workers share has a p-value in the 5%-10% range, meaning
that it is not clear whether the coefficients are to be considered different or not across groups. We
can think that a firm endowed with a high share of temporary workers always exhibits lower
productivity growth.
Finally, the capital-labor ratio coefficient is usually significant and positive for the innovative firms,
its point-wise estimate is almost three times bigger (.26) than that of non-innovative firms (around
.07 in the case of product innovative firms). The three Wald tests of parameter stability always
18