We have argued before that the procyclical bias in fiscal policy is mainly due to reversals
in taxes. We indeed confirm the procyclicality of revenues as in all countries, elasticities
are significantly larger than the corresponding elasticities from OECD (see also Table
5.1).101 This is especially pronounced in the nineties in all countries, with the exception of
Germany. The changes over decades are quite outspoken and hide quite some adjustments
in tax systems. Only in Germany is the response of revenues procyclical in all sub-
samples. For France, Portugal and Spain, the elasticities in the seventies are not
significant. This must be related to the development of tax systems in the latter two
countries; the result for France seems more puzzling.102
These results show that latent policy pressures on spending or revenue bring about
adjustments that usually reverse the effects of automatic stabilisers. The ‘actual’
elasticities incorporate all cyclical reactions, coming from the automatic adjustments via
the underlying tax and spending structure and systematic interventions of fiscal
policymakers. If we choose to impose the ‘actual’ elasticity in the VAR model (5.5), the
interpretation of the fiscal shock is one that includes all discretionary interventions. The
drawback of the approach is that our ‘cyclical’ shock is a mongrel reaction to economic
conditions, in which we cannot tell apart the importance of systematic policy and the
economic cycle. The difference in the structural indicator - obtained with the OECD
elasticities - can then be attributed to the procyclical bias in fiscal policy.
Table 5.8 summarises the elasticities that we have taken from Table 5.7 for the entire
sample period for re-estimating the VAR. Figure 5.6 compares the structural indicator.
We find convergence to the same solution as in the basic case: there are only some
marginal differences for the case of Portugal.
|
Table 5.8 Elasticities Imposed on Equation 5.5 | ||||
|
France |
______Germany |
Portugal |
Spain______ | |
|
Total expenditure, γ |
0.32 |
1.04 |
0.67 |
0.03 |
|
Total revenue, α |
1.73 |
1.47 |
1.58 |
1.36 |
|
Source: author's calculations. | ||||
What does the insensitivity of the results to assumptions on the budget elasticities tell us?
The forecast error variance decomposition reveals nearly equivalent roles for demand
effects of fiscal policies and supply shocks in Germany and France, whereas supply
shocks tend to dominate in Spain and Portugal. If we recover nearly similar fiscal policy
shocks whether correcting for automatic stabilisers, setting them to zero or taking the
systematic variation in fiscal policy into account, this is due to the little importance of
cyclical economic shocks. This does not mean that the automatic stabilisers are irrelevant.
The stabilising effects of the structure of the spending and taxation system will still work
101
102
If the government decides to raise tax rates in economic crises, this leads to a stronger than expected
reaction of revenues in the following economic boom.
The time variation in elasticities is also apparent from a recursive regression of 5.7. Coefficient plots
are summarised in Appendix 5.C and further documents some of the problems with constant
elasticities. We have not reported the elasticities of interest payments and investment, as these
coefficients are much more volatile than those of other budget items. There are relevant breaks
associated with major shifts in fiscal policy (e.g. German Reunification, democracy in Portugal and
Spain). For most spending categories, we remark a modest decline over time in Germany and a more
outspoken one in France, while changes are minor for most revenue categories. Portugal and Spain
have seen a large rise in elasticities of all items, owing to the expansion of their welfare states. This
rise has pushed elasticities even above those in Germany and France.
152