positive output effects. We do not find the typical result of small positive Keynesian
effects on output in all countries. In Germany and Spain, a typical Keynesian response
would follow upon demand boosting deficits. In France and Portugal on the other hand,
fiscal contractions would lead to positive short-term effects on output instead. Such
different responses likely depend on the composition of the fiscal adjustment or other
structural parameters in the respective economies, but cannot be further examined in the
current model.
The different responses of spending and revenues to both economic shocks might indicate
a delicate issue in the identification of policy. If fiscal policy reacts in a systematic way to
economic shocks by changing its discretionary use of spending and/or revenues, this
simultaneity blurs the distinction between the economic and the fiscal shock. This might
be the case in France and Spain where tax revenues decrease after positive temporary
output shocks, for example. But another indication is given by the rise in spending in
economic booms in Germany or Portugal. It indicates policies that react in a discretionary
way so as to repeal the use of automatic stabilisers. The fiscal indicator captures these
policy biases. Our discussion will show how important this policy bias is for
understanding fiscal trends in EU countries.
What does this imply for the contribution of fiscal policies to output variation (Figure
5.4)? Supply shocks account for at least 50% of total variance in output at all horizons,
and this goes up to 90% in Portugal and Spain. For the latter countries, this was perhaps
to be expected given their strong economic growth over the last two decades. Most of the
variation in output is thus caused by productivity shocks even at short horizons. As we do
not separately identify private and public supply shocks, we cannot really quantify the
relative magnitude of both channels. But as pointed out above, we think that productive
spending or revenues has contributed to some extent to the variance of output. The
demand effects of fiscal policy in France and Germany are at least as large as those of
supply effects. In Portugal or Spain instead, only a minor role is played by discretionary
fiscal policy. The contribution of cyclical fluctuations to variations in output is negligible,
as was to be expected from the results on the impulse responses.
What factors can account for these results? The large role played by fiscal policy in
explaining output variation is not inconsistent with previous findings in the literature for
large EU countries (De Arcangelis and Lamartina, 2004), but seems on the higher side of
the range usually found. If we take the result at face value, it would suggest that the
temporary demand effects of fiscal policy are probably much larger than the supply
effects in the long-term. This would imply that both RBC and New Keynesian models are
missing some aspects of fiscal transmission. But as we cannot precisely quantify the
importance of the latter shocks, we would not want to claim validation of any of the
theoretical models with our approach. This result nevertheless reveals that models of
fiscal policy need to attribute important roles to both demand and supply side effects.
139