|
coeff. |
std.dev. |
t-test |
p-value | |||
|
pr{⅛ I Vi,t-ι,Xi,t} |
βιo |
Constant |
-3.798 |
(∏29^ |
-29.51 |
(ШГ |
|
βl2 |
ylt-ι |
0.469 |
0.125 |
3.76 |
0.00 | |
|
βl5 |
sizβi |
(L285^^ |
(H)43^ |
ÔT |
0.00 | |
|
βl6 |
t ~ |
(∏3SΓ |
(H∏Γ |
12.48 |
0.00 | |
|
-¾. |
Tj,t______________ |
-0.003 |
0.009 |
-0.30 |
0.38 | |
|
pr{⅛ I Vi,t-ι,Xi,t} |
β20 |
Constant |
-3.562 |
0.192 |
-18.56 |
0.00 |
|
β2i |
yi-ι ^ |
0.198 |
0.109 |
1.81 |
0.03 | |
|
β25 |
sizβi |
0.252 |
0.057 |
4.41 |
0.00 | |
|
β26 |
t ~ |
(≡2^ |
(HΠ8- |
4^6Γ |
0.00 | |
|
β27 |
Tj,t______________ |
-0.006 |
0.014 |
-0.41 |
0.34 | |
|
P__ |
Correlation |
0.246 |
0.081 |
3.05 |
0.00 | |
|
Observations |
9853 | |||||
|
Loglikelihood |
_______________-1475.488 | |||||
Table 5: Estimates of the Unsaturated Model
and FMS: it has a positive sign, confirming the presence of an effect of size
on the returns from adoption. Calendar time is also highly significant, because
of its positive correlation with the performance of both CADCAM and FMS,
and its negative correlation with prices. Duration of non adoption results in-
significant, but this is probably due to a problem of misspecification, since a
more complete study on the same dataset evidences a negative and moderately
significant impact of this variable on the probability of adoption of FMS, while
no effect is detected on CADCAM.
Based on Table 5, Wald type поп-causality tests may be done by simply
analyzing the t — test for the parameters βi2, β2ι and p. βi2 is positive and
significant (the hypothesis of Granger поп-causality is rejected), which suggest a
positive effect of adoption of FMS on the following adoption of CADCAM: this
has a simple economic interpretation and confirms the presence of an interaction
of the two technologies. On the other hand, β2ι is positive but non significant,
which means that the Wald test accepts the hypothesis that CADCAM does
not Granger cause FMS (У1 → У2). The correlation between the error terms
of the latent regressions relative to CADCAM and FMS (p) is positive and highly
significant: this implies a positive interaction in the simultaneous adoption of
the two technologies. It is worth noticing that the results of поп-causality tests
depend on the information set, and therefore one might think that the evidence
presented here depends on the very limited information supplied to the model.
Actually, the results of the поп-causality analysis are substantially unchanged
even when the variables conditioned upon are all those included in Colombo
and Mosconi (1995).
Let us now illustrate a more complete поп-causality analysis, including some
interesting joint hypotheses. Testing is based on likelihood ratio tests; Wald
tests have been also computed, getting essentially identical results. For both the
saturated and the unsaturated models, four restrictions on Hjj are considered:
24
More intriguing information
1. Developmental Robots - A New Paradigm2. Ein pragmatisierter Kalkul des naturlichen Schlieβens nebst Metatheorie
3. The Prohibition of the Proposed Springer-ProSiebenSat.1-Merger: How much Economics in German Merger Control?
4. Problems of operationalizing the concept of a cost-of-living index
5. The name is absent
6. The name is absent
7. SME'S SUPPORT AND REGIONAL POLICY IN EU - THE NORTE-LITORAL PORTUGUESE EXPERIENCE
8. 5th and 8th grade pupils’ and teachers’ perceptions of the relationships between teaching methods, classroom ethos, and positive affective attitudes towards learning mathematics in Japan
9. The name is absent
10. The Functions of Postpartum Depression