The results of the estimation of equation (30), reported in Table 5, show that the price
level does not seem to have a statistically significant relation with the government
revenues. This is true for the pooled regression, fixed effects and random effects
versions, even though now the fixed effects model is not statistically better when
compared against the other two versions. Furthermore, such a relation, even if
significant at the 10 per cent level, is nevertheless positive, and not negative, as one
would expect if the price level adjusted upwards after a decrease in public revenues.
Therefore, one could hardly decide, with this evidence, for the validation of the FTPL in
the EU-15 countries.
Table 5 - Estimation of equation (30), dependent variable: average annual change
______________________of the price deflator______________________
Variable |
Pooled |
Fixed effects |
Random |
Constant (λi) |
0,650 |
-0,704 | |
ωP (Pit-1) |
0,892 |
0,850 |
0,878 |
Or (R⅛-1) |
0,175* |
0,187* (1,800) |
0,179* (1,774) |
R'' |
0,8029 |
0,8008 |
0,8029 |
F α test |
0,683 | ||
Hausman b test |
5.601 ** | ||
DW_____ |
1,789 |
1,754 |
1,762 |
The t statistics are in parentheses.
a - The degrees of freedom for the F statistic are in parentheses; the statistic tests the fixed
effects model against the pooled regression model, where the autonomous term is the same for
all countries, which is the null hypothesis.
b - The statistic has a Chi-square distribution (the degrees of freedom are in parentheses); the
Hausman statistic tests the fixed effects model against the random effects, which is here the null
hypothesis.
* - Statistically significant at the 10 per cent level.
** - Statistically significant (only) at the 10 per cent level, the null hypothesis is rejected
(random effects model), that is, one rejects the hypothesis that the autonomous terms in each
country are not correlated with the independent explanatory variables (in this case the random
effects model does not produce unbiased and consistent estimators).
29