response to debt through the inclusion of ad hoc dummy effects associated with specific
events along the process of economic and monetary integration in the European Union.
This analysis does not change the qualitative baseline result, but reveals the existence of a
Maastricht/euro-launching positive correction in the response to debt in many countries,
and the existence of a “Maastricht fatigue” in some of them, indicating time variation in
the response of governments to debt accumulation.
Another main concern of this chapter is to provide an alternative to ad hoc dummies in
order to investigate this time variation in the response to debt accumulation from a more
systematic perspective. In this respect, we argue that a Bayesian approach provides a
sensible endogenous updating scheme in our small sample context, and we estimate a
Bayesian version of the baseline model for three different sub-samples: 1977-1991, 1992-
1999, 2000-2005. We find that the posterior distribution has tended to shift across sub-
samples in most countries, which supports the hypothesis of a time varying response to
debt. The comparison of posteriors provides an explicit characterization of this time
variation, and indicates that a positive posterior mean for the response to debt
accumulation is pervasive. This enhances the robustness of our results.
From a policy perspective, we see two messages from our analysis. First, if sustainability
is the issue, there is no clear reason for concern. The evidence for last 30 years suggests
that, with the exception of Greece, EU15 governments have tended to apply a fluctuating
but generally positive primary surplus adjustment in response to debt accumulation. This
is sufficient to guarantee the sustainability of fiscal policy.12 Second, the posterior
distribution from our Bayesian analysis provides a sensible, easy-to-update indicator for
the evolution of the sustainability position of countries. A country with a posterior
centred on a non-positive response for a given number of periods could be called for a
fiscal correction on sustainability grounds.
Finally, a comment on follow-up research. A natural direction for further work in the
issue of sustainability is to try to specify and estimate a model with time varying
coefficients. This may require longer samples. A successful attempt though would
generate time series for the response of the primary surplus to debt accumulation, thus
providing the raw material that would allow moving from the objective of testing
sustainability, the focus of this chapter, to the objective of identifying its potential
determinants.
Although the size of response is also relevant in order to exclude explosive debt-to-GDP ratios.
46