Non-causality in Bivariate Binary Panel Data

Table 6: Likelihood of the Estimated Models

Table 7: LR Non-Causality Tests for the Unsaturated Model

H_lw2 : Y¹ → Y²,

H_lw2 -.Y² → Y¹,

H_lw2 -.Y¹ → Y², Y² → Y¹,

Я1_±2 :uɪ →Y²,Y² →Y¹,Y¹ φY².

The maximized loglikelihood, together with the corresponding number of pa-
rameters, is reported in Table 6 for each of the estimated models.

It is straightforward to note that likelihood ratio tests fail to detect, at usual
significance level, any difference between a saturated model and the correspond-
ing unsaturated one. It is therefore not surprising that also the non-causality
tests yield about the same results. Table 7 reports the tests for the unsatu-
rated model. The results support the economic consideration that the previous
adoption of an FMS rises the probability of adoption of a CADCAM, while in
the opposite direction the evidence is much weaker, and statistically non sig-
nificant. In fact, the economic intuition suggests that CADCAM is a powerful
design tool even without an FMS. The strong simultaneous dependence suggest
that the decision of joint simultaneous adoption occurs more often than what
would be expected if the two decisions were taken independently.

8 Conclusions

In this paper we make a step towards rendering an important tool of applied
macroeconometric analysis, such as Granger поп-causality, available and oper-
ational for those situations in which the processes involved in the analysis are

25

<<   <   22   23   24   25   26   27   28   >   >>