euro at the end of 1995. As a result, a non-significant reaction in Greece and Austria
became significant during the nineties, thus fulfilling the sustainability condition (2.6). In
the case of Spain, a relatively weak response became much stronger after 1996. In France,
the positive and relatively large response of the primary balance to debt was halved after
1993, but still leaving it to a size comparable to that estimated for other countries (around
0.05). In Japan, the change in fiscal policy that occurred after the crisis of the early
nineties has totally offset the secular positive reaction of the primary surplus to debt.
Interestingly, in a number of countries, namely Germany, Greece, France and Italy, there
is some evidence of a sort of EMU fatigue, which has lowered the reaction of the primary
surplus to debt during the 2000s. The reduction appears sizeable in Germany, as well as
in France, where it added to that taking place after 1993. The reduction has also been
important in Greece, where the positive reaction of 0.03 that prevailed during the nineties
left way to the behaviour already observed in the seventies and the eighties.
All in all, after controlling for breaks in the baseline model (Table 2.3), Table 2.6
confirms that the fulfilment of the sustainability condition given by (2.6) has been general
across the sample.10 Greece and Finland seem to be the only outstanding exceptions.
However, in the latter case, one should bear in mind that the events of the period 1990-
1992, when the effects of the fall of the former USSR and the 1992 crisis overlapped
prompting a dramatic increase of debt levels (see Section 2.3), made the estimation of
reaction function for this country very difficult. In addition, the primary surplus has
started to fall only recently while its debt has been on a downward path since 1993, which
does not point precisely to an unsustainable fiscal behaviour. On the other hand, we
should not forget that our requirement for sustainability is a sufficient but not a necessary
condition.
2.5 A Closer Look at the Response of Primary Surplus to Debt
The analysis in Section 2.4 provides robust evidence supporting the conclusion that most
EU countries can be characterized as having a positive response of the primary surplus to
debt accumulation over the full sample period. It also allows to identify some significant
shifts in the response of some countries as the economic and monetary integration process
in Europe has evolved, indicating the existence of some degree of time variation in that
response.
However, the analysis of last section does not allow tracing with some precision that
degree of time variation. This is so because it cannot provide relatively precise estimates
of the value of the response in small sub-samples, due to the lack of degrees of freedom.
It provides estimates of the average response for the overall sample period, which is then
permanently corrected with ad hoc dummy effects after specific sample events. But it
does not provide a systematic updating mechanism to obtain explicit probability
distributions of the primary surplus response to debt for different (small) sub-sample
periods. Explicit characterization of these distributions would be helpful for a systematic
investigation of the statistical evolution of time variation over the sample.
Implicit in our analysis is the assumption that debt and primary surplus have the same order of
integration. Appendix 2.B reports unit root test results which fail to display unambiguous statistical
evidence in this sense. However, our sample size is too small for this matter and we take this as
inconclusive evidence.
40