used for this purpose policy induced temporary factors (box B) are not netted out. In fact,
this distinction is increasingly recognised and for example the Commission now regularly
net out "one-off budget measures" from the CAB and labels this as the “structural budget
balance.”
Second, the “fiscal stance” relates to whether the budget balance, over and beyond the
impact of the automatic stabilisers, contribute or not to smooth the cycle. This would
imply netting out the cyclical economy induced component and look at boxes B+C+D.
Using the authors model the "discretionary component" would probably be used to assess
the fiscal stance. However, the change in the CAB would be better designed for this
purpose. While it is straightforward that boxes B and D should be included in a measure
of the fiscal stance it may be less clear with box C which takes into account economy
induced permanent changes. Ultimately it depends on the precise question to be analysed.
Lastly, if the interest is to identify the impact of discretionary policy decisions, then
temporary and permanent economy-induced budget components should not be taken into
account (thus look at B+D). Here it is worth noting that if the CAB is used for this
purpose it would include also economy induced permanent changes (box C) while the
discretionary component in the authors model would not capture policy induced
permanent changes (box D). The overall point here is that it is important to be precise in
what is the objective of the analysis and what the indicator used actually captures.
Conceptually at least, the authors "core component" would be better than the CAB when
assessing the structural budget balance while the change in the CAB would be best
designed to capture the "fiscal stance" as understood here. To study the "discretionary
policy" response of the government the model looks better designed from a conceptual
perspective than the change in the CAB.
It is also worth looking a bit closer at some aspects related to what is actually meant by
"discretionary policy" when using budget indicators.66. One issue relates to budget rules
across layers of government. For example, say that a local government (where most of
government consumption takes place) runs deficits to cover for higher expenditures,
albeit they are forbidden to borrow, but that they are bailed out by the central level
through additional grants. This could be regarded as a discretionary measure by the
government (box B above), given that a separate decision was taken to this end, but it
could also be regarded as “semi-automatic” if it is common practice (box A above).
Clearly, how it is counted may make a difference in the policy assessment. Another issue
relates to the benchmark used for what is “neutral policy” and then in particular the
indexation of expenditures. CAB calculations based on trend GDP estimates use as
implicit benchmark that expenditures develop in line with potential GDP. If some
expenditures, such as welfare expenditures, are instead indexed to nominal GDP the
difference between the two growth rates would not be included in the "cyclical part" and
thus be counted as non-cyclical. In fact, to the extent that expenditures are actually
indexed, to say inflation, these items could be labelled “automatic destabilisers”! Overall,
it is the non-indexation of expenditures which provides most of the cyclical stabilisation
from the budget.
My last comment relates to the budget balance itself: observable and the base for most
budget indicators. Further work to decompose the nominal balance in sub components
allowing better to analyse different budget behaviour is of course very useful to make
See also Boije and Fischer (2006).
117