*1∣ Y I *1∣ K I .O. ~T— . I Temporary ʌ ,
δ 2lnl7l =αλ 2lnl^∏ + β+γAgei,2001+ μ∖—~τ—- + εi (4)
∖ L J i,2001-03 ∖ L ) i,2001-03 ∖ L J i,2001
The dependent variable is the 2001-03 growth rate of labor productivity for firm i calculated at time
t = 2003, with respect to 2001. Age is calculated as the average age of the board members and
managers when they were appointed.1 Temporary/L is the share of workers in the firm operating on
a temporary contract (full time + part time) in 2001. In the regressions we control also for twenty-
one sector dummies,2 four geographical macro areas,3 size dummies for small, medium and large
firms and firm membership in a group. Size is measured following the European Commission
definition: firms with less than 50 workers are “Small”; firms with more than 50 but less than or
equal 250 workers are “Medium”; firms with more than 250 employees are “Large”.
Unfortunately, the IX Capitalia/Unicredit survey includes three years of observation only for a
limited subset of our variables of interest. As to the determinants of innovation, in particular, we
only have data for 2001, the initial year in our sample, and not for the three years between 2001 and
2003.4 Hence specification (4) is estimated by regressing the growth rate of labor productivity
(hence the long difference of log productivity levels) on explanatory variables measured at a point
in time, therefore within a cross sectional framework. Yet first differencing the log levels of labor
productivity and the capital-labor ratio allows us to get rid of some of the unobserved heterogeneity
between firms that represents the most obvious source of simultaneity bias.
As a second step in our empirical analysis, we run a Chow test of parameter instability on
specification (4) to check whether there are significant asymmetries between innovative and non-
innovative firms. It might be that because we expect the parameters (β,γ,μ) to differ between
1 As a robustness check, we have also substituted as a regressor the average age of the board members with their
seniority, i.e. the firm-board-average number of years in the board in 2001 (the initial year of our sample). This measure
may however contain substantial measurement error given that we do not know when board members have quit or
changed their role within the board before 2003. We have thus chosen not to report these results, which are however
available upon request.
2 The sector breakdown is based on the Ateco2007 classification of Italy’s industries, in turn equivalent to the NACE
rev.2 European code.
3 Macro areas are defined by the Italian National Institute of Statistics (ISTAT) which groups Italian regions into 4
areas: North West (Lombardy, Piedmont, Liguria), North East (Veneto, Trentino Alto Adige, Friuli Venezia Giulia,
Emilia Romagna), Centre (Lazio, Umbria, Marche, Tuscany), South and Islands (Campania, Apulia, Abruzzo, Molise,
Basilicata, Calabria, Sicily, Sardinia).
4 If merged with the data from a previous survey, the IXth survey plus AIDA reduces to a tiny sample of firms.