instead be real output less (real) materials and services. This implies that materials and services are
no longer an argument in the production function.
Subject to these preliminary remarks and assumptions, in each period t, labor productivity (in logs)
for firm i at time t may be decomposed as follows:
ln(^ / Llt) = ln(Au) + βκ ln(K . / Lit) (1)
Equation (1) stems from a production function where (the log of) value added Y is a log-linear
function of the capital labor ratio K/L and the efficiency parameter A. In turn, A is a function of
time and innovation as follows:
ln(Ait ) = α * t * INNOVATION + error (2)
The growth rate of the technological parameter is thus a linear function of INNOVATION. Under
(1) and (2), the log difference (the growth rate) of labor productivity is a linear function of the
growth rate of the capital-labor ratio and of the determinants of innovation.
In turn we assume that INNOVATION is a linear function of a few variables of interest including
our preferred proxies for experience, i.e. firm-averaged managerial age and the share of part-time
temporary workers in each firm, and other determinants of the decision to innovate (whether a firm
undertakes R&D spending, the share of R&D workers in the total firm’s labor force; cash flows)
plus an array of regional, size and industry dummies, each affecting (the log of) A through a
separate parameter.
Leaving aside the other determinants of innovation for expositional purpose, INNOVATION may
be seen as a function of managerial experience and business schooling capital as follows:
INNOVATIONi = a Ei + b Si (3)
where E is experience and S is managerial capital formally accumulated going to the business
school with S = T - E. The variable “E” is the number of years a manager has spent doing her job
inside or outside the firm. The variable “S” is the managerial capital accumulated at the business
school by the manager under the time constraint T=E+S, i.e. a manager either goes to the business