especially in countries where the budget deficit is not far enough from its 3% ceiling and an
asymmetric use for countries with an external competitiveness shortfall.
3.3. The evidence
Some of these presumptions have been tested. Looking at Eurozone countries, Gali and
Perotti (2003), Fatas and Mihov (2001, 2003), and Wyplosz (2005) have found that, in most
cases fiscal policy has been acyclical, sometimes even procyclical. They also report that the
constraints imposed by the monetary union, the convergence criteria of the “Maastricht years”
1992-98 and the SGP since 1999, have led to somewhat less policy activism and, as a
consequence, to less procyclical policies. Fatas and Mihov (2001, 2003) further document that
the SGP constraints seem to have mitigated the various influences that are believed to distort
the use of the fiscal policy instrument.
These estimates are based on the early experience of monetary union. We revisit them with
data that extend to 2006, see the data appendix. We adopt the formulation proposed by Gali
and Perotti (2003):
(4) dt =c+βbEt-1yt+βaEt-1yt+γbt-1+ρdt-1+ut
where dt is the annual budget deficit, Et-1yt is the expected output gap and bt is the public debt
(dt and bt are expressed as percent of GDP). The coefficients βb and βa correspond to the
period before and after 1992 onward; the break is introduced to test whether the restrictions
adopted in the Maastricht treaty have affected the cyclical use of fiscal policy. An alternative
is to break the sample in 1999, or even to consider three subperiods, separating out the
Maastricht from the monetary union years, but the data support the break as indicated. The
expected output gap is estimated as in Gali and Perotti (2003) by replacing Et-1yt with yt and
instrumenting it with the US and Japanese output gaps and the lagged output gap yt-1. A
countercyclical use of fiscal policy implies that βb < 0 and βa < 0 .The assumption that the
adoption of the common currency has fostered a substitution of fiscal for monetary policy as
the macroeconomic stabilization tool implies that βb > βa .
We also use this framework to test the other hypotheses presented above. To test whether the
SGP has an asymmetric effect over the cycle, limiting the countercyclical use of fiscal policy
in downswings, we replace βaEt-1 yt with βaEt-1 yt + βa-Et-1 yt- where yt- is the output
gap when it is negative and zero when it is positive. The hypothesis implies that βa- > 0 .
We have also noted that fiscal policy could be used as a substitute for lost monetary policy
when dealing with external competitiveness. One hypothesis is that, independently of the
cyclical position already captured with Et-1yt, fiscal policy is expansionary when external
competitiveness is declining. Another hypothesis is that fiscal policy is instead used to reduce
costs and regain competitiveness by being restrictive. We can test which hypothesis is borne
out by adding:
θbX +θaX
where X is a measure of external competitiveness and θb and θa correspond to before and
after 1992. When fiscal policy is used to compress costs when competitiveness is low, we
expect to find θi > 0 . If instead fiscal policy is used to offset the demand effects of poor
competitiveness, we should have θi < 0 . If the Maastricht Treaty and the creation of Euro
area have reinforced the use of fiscal policy as a countercyclical tool, we should find
θa < θb ≤ 0 . If instead fiscal policy was initially used as a demand management tool before