Yet declining productivity is also - slightly less robustly - associated to such managerial features as
the average age of the managers running the companies. Our estimates separately run for innovative
and non-innovative companies indicate that an older age of managers is associated with lower
productivity in innovative firms and with higher productivity in non-innovative firms. This is
consistent with common sense that suggests a more positive role of experience in firms with
relatively standardized and stable business practices, while old age is presumably more damaging
for innovative firms that would be supposed to swiftly adopt new technologies as they become
available. This is presumably more tightly correlated with schooling and less with experience on the
job. Our two-stage model results, while broadly confirming the robustness of the partial correlation
between the share of temporary workers and productivity, shows that both product and process
innovation are positively related with productivity growth. For the innovative firms age is
negatively correlated with productivity growth and positively correlated or at times uncorrelated
within the sample of non innovative firms, with little variation across estimation methods.
Our paper is structured as follows. In section 2, we discuss why and how experience should matter
for innovation and productivity. In section 3 we describe the paper’s conceptual underpinnings and
estimation strategy. In section 4 we describe the main features of our data set. In section 5 we
present our main results and some extensions. Section 6 concludes.
2. Conceptual framework
We consider a constant-returns-to-scale value-added production function. The full-fledged
production function underlying equation (1) below would have (real) output on the left-hand side
and capital, labor, intermediate inputs and services on the right-hand side. This would allow us to
differentially treat the substitutability of such inputs with respect to capital and labor. Specifying the
production function in terms of value added, however, lessens the endogenous input choice problem
that plagues the estimates of production functions in general. Under the (untested) assumption of
separability between the value added and the intermediates functions, the dependent variable may