The initially reported increase in deficit between 2002 and 2003 (from 1.4 percent to 1.7
percent of GDP) was in line with that observed for the change in debt, the latter rising
from 5.6 percent of GDP in 2002 to 5.9 percent in 2003. However, the level of the two
indicators was markedly different (Panel A of Figure 6.3). Panel B shows how revisions
have completely cancelled the 2003 discrepancy and significantly reduced those for
previous years.
6.4 Fiscal Rules and Fiscal Gimmickry: a Simple Model
An econometric analysis of SFA in EU member countries was first provided by von
Hagen and Wolff (2005). Their paper refers to the theoretical framework developed in
von Hagen and Harden (1995, 1996) and Milesi-Ferretti (2003), where governments have
an incentive to circumvent fiscal rules by hiding the budgetary implications of fiscal
policies in less visible accounting items (that is, in the SFA). The likelihood of this type
of window dressing decreases with the costs associated with detection. The authors argue
that binding deficit rules were introduced only with the start up of the European
Economic Monetary Union (EMU) - i.e. the SGP - and therefore focus their analysis on
differences in the correlation between reported deficits and SFA before and after EMU.
They find no such correlation before 1998, but a negative one (large and significant)
thereafter, suggesting that SFA were in fact substituting for other transactions which
would have had an impact on deficits.
Buti et al. (2007) develop a model where total SFA is split into two components (one that
can be used to reduce reported deficits and the other to impact debt figures). In this way
they can separately analyze the interaction between each of the Maastricht fiscal rules and
fiscal gimmickry. They assume that governments minimize a quadratic loss function
whose arguments are the deviation of output from its optimal level (influenced by the
“true” deficit), deviations of reported deficit and debt from the respective fiscal rule, and
the size of accounting gimmicks. The model suggests that both the deficit-specific and the
debt-specific components of SFA are positively related to the “true deficit”, and that only
the debt-specific component also depends on the debt level (though the sign of the
relation is ambiguous ex ante). The empirical results are partly in line with the theoretical
predictions of the model.117 Notably, the authors find that the introduction of the SGP had
an (increasing) impact on the deficit-specific component of the SFA, but none on the
debt-specific component.
In this section, following Buti et al. (2007) we provide separate econometric analysis of
deficit-specific and debt-specific SFA components. However, we refer to a different
model as the basis for our estimating equations. We assume that governments derive
utility (U) from running primary deficits (p): U=U(p). This can be justified either by
assuming that governments are short sighted and only care about the short-term output
gains that can be attained through higher deficits, or by reference to the political gain
directly attainable by increasing transfers targeted to specific groups. In either case, the
The authors find no statistically significant relationship between the two components of SFA and
“true deficits”. They find evidence of a positive relationship between reported deficits and deficit-
specific SFA and of a negative relationship between reported deficits and debt-specific SFA. Only
debt-specific SFA are found to be affected (negatively) by the debt level.
172