Table 6. Second stage results for system (9’), OLS and 2SLS.
Dependent |
OLS |
2SLS |
OLS |
2SLS | |||||
K ∆2 ln — 2 L Lit |
Coeff |
Robust St.E. |
Coeff |
Robust St.E. |
Coeff |
Robust St.E. |
Coeff |
Robust St.E. | |
Innovative firms |
Non Innovative firms | ||||||||
α1 (K/L ratio) |
0.256*** |
0.0265 |
0.199*** |
0.0302 |
«2 |
0.087*** |
0.0212 |
0.085*** |
0.0239 |
β1 (constant) |
20.164*** |
3.204 |
20.46*** |
3.146 |
β2 |
-6.68 |
7.755 |
-6.67 |
7.698 |
Yi (age) |
-0.307*** |
0.0465 |
-0.313*** |
0.0453 |
Y2 |
0.061 |
0.1375 |
0.061 |
0.1366 |
μ1 (temporary |
-0.114*** |
0.023 |
-0.123*** |
0.0234 |
μ2 |
-0.273** |
0.120 |
-0.272** |
0.1194 |
-σεiω |
2.597* |
1.526 |
3.659** |
1.528 |
σε2ω |
-1.557 |
3.152 |
-1.545 |
3.1343 |
R2 |
0.215 |
0.099 |
R2 |
0.208 |
0.099 | ||||
Sargan test |
3.429 |
.413 | |||||||
N |
5859 |
5859 |
N |
1914 |
1914 | ||||
Note. All regressions include size, sector, area and group dummies. The first stage refers to column 1 and 2 of
Table 5. The 2SLS method instruments the growth rate of the capital stock per worker with the lagged K/L in
levels and the lagged investment intensity.
As a robustness check we replicate the estimates of equation (4) for the sub-sample of managers
who are supposed to take decisions within the board. We select those firms whose board contains
competence-specific managers, who are supposed to influence the decision to introduce innovations
in the firm, eventually. The sign and size of the main coefficients are confirmed. We omit reporting
them here for brevity.
We finally implement the Maximum Likelihood Endogenous Switching model which allows us to
obtain consistent and efficient estimates.
22