Table 7. Second stage results for system (9’), Maximum Likelihood efficient method.
|
Dependent |
ML |
/V ML (predict k ) |
ML |
/V ML (predict k ) | |||||
|
K ∆7 ln — 2 L Lit |
Coeff |
Robust St. E. |
Coeff |
Robust Std. Err. |
Coeff |
Robust St. E. |
Coeff |
Robust St. E. | |
|
Innovative firms |
Non Innovative firms | ||||||||
|
αι |
.2588*** |
.0244 |
.1967*** |
.0382 |
α2 |
.0199 |
.0195 |
-.0004 |
.0252 |
|
21______ |
17.00*** |
3.343 |
18.17*** |
3.381 |
β2_______ |
-6.658 |
10.85 |
-6.414 |
10.023 |
|
Yi |
-.4028*** |
.056 |
-.3862*** |
.0591 |
Y2 |
-.0105 |
.1473 |
-.0051 |
.1462 |
|
μι |
-.1878*** |
.0288 |
-.2172*** |
.0308 |
μ2 |
-.242** |
.1174 |
-.2369** |
.1167 |
|
σι |
28.719*** |
.8895 |
30.51*** |
1.096 |
σ2 |
29.941*** |
.2622 |
29.93*** |
.1986 |
|
Pi |
.6917 |
.0936 |
.6421 |
.1211 |
P2 |
.1126 |
.1385 |
.1039 |
.1179 |
|
Mills ratio |
.4439** |
.2767 |
.444** |
.2821 |
1.364*** |
.5181 |
1.368*** |
.5282 | |
|
Condition |
Exogenous K |
Endogenous K | |||||||
|
Wald χ2(i) |
22.86 [0.000] |
13.95 | |||||||
|
P(Inno) |
.7356 |
.1785 |
.735 |
.1816 | |||||
|
N |
7773 |
7773 | |||||||
Note. All regressions include size, area and sector dummies. Columns 3 and 4 show ML with predicted value of the
growth rate of capital stock for innovative firms. Columns 7 and 8 replicate ML with predicted value for the non
innovative cluster. Wald Chi-square test is testing for the independence of the residuals in system (8). P(Inno) is the
predicted probability of being in the innovative regime in the first stage.
8. Conclusions
In this paper we exploited data from a sample of some eight thousands innovative and non-
innovative firm-observations to describe the pattern of correlation between experience, innovation
and productivity growth during the recent period of serious productivity slowdown in the Italian
24