has instead kept the average age of those in charge of leading Italian companies unchanged, and this
missing change (and “excess experience”) is alleged to have lowered the propensity to innovate and
the productivity performance of Italian firms. Gordon and Dew-Becker have not contrasted their
ideas with micro data, while Bandiera, Guiso, Prat and Sadun have not looked at the interaction of
labor market reform with the productivity and innovation counterpart of managerial practices. So
there is room for comparing a streamlined version of the two views with company data. This is the
main goal of our paper. More generally we believe that this combination of events and structural
features is helpful for learning on the relation between experience, innovation and productivity.
In our study we employ a firm-level data set to separately analyze at the role of experience on the
side of workers and the side of managers. Our preferred measure of experience on the workers’ side
is the average share of temporary workers in total employment. The increased presence of
temporary workers in the Italian labor force has been a novelty of the late 1990s, which is portrayed
in our company data. Our preferred measure of experience on the manager side is the average age of
managers and board members in the Italian firms.
In doing so, we are confronted with (and we thus explicitly tackle) a few statistical hurdles, the first
of which is the fundamentally cross-sectional nature of our data. The use of (long) differences for
productivity growth (as opposed to productivity levels as most other studies such as Hall, Lotti and
Mairesse (2007) allows us to lessen the simultaneity bias that would originate from regressing log
levels of firm performance onto our variables of interest, such as managerial age or the share of
temporary workers. Secondly, our preferred set of estimates is the result of a first stage where a
regime switch from being non-innovative to innovative or the reverse may occur or not and a
second stage.
Our evidence indicates that innovation and productivity growth was particularly low in firms with
disproportionately high shares of temporary workers to start with. This result is robust to all
changes of specifications.