Figure 5. Aggregate structural policy stance indicator
29. As in Duval (2005) we consider 21 countries for which we have annual observations for the
period 1985-2003. Using this dataset, we estimated the following partial-adjustment relationship:
(1) PRIit=λPRIit-1+αSTRit+βΔSTRit+∑γkCONikt-1+δi+εit
k
30. In this relationship PRIit is the level of cyclically-adjusted primary expenditure as a per cent of
GDP in country i in year t. STRit is the overall structural policy stance. It is included to capture the long-run
effect of the structural policy stance on public expenditure. The term ΔSTRit is the change in the structural
policy stance indicator which serves to capture any upfront budgetary cost of structural reform - its
expected sign is negative as predicted by theoretical models such as that reported in Beetsma and Debrun
(2005). CONit-1 is a vector of control variables, δi are country fixed effects and εit is the normally distributed
residual. The method used is ordinary least squares.
31. Following Martinez-Mongay (2002), four control variables have been considered:
• Per capita gross national income at 2000 purchasing power parities. This captures
“Wagner’s law”, which predicts that high-income countries will exhibit higher shares of
public spending in GDP than low-income countries owing to a change in preferences in
favour of public goods and services such as health care, education and social services.
The expected sign is positive.
• The dependency ratio. Ageing puts pressure on notably health care and pension
expenditure, hence a priori one expects public outlays to be higher in countries that
portray a high dependency ratio (measured by the share of people older than 65 in the
total population). The expected sign is again positive.
• Trade openness (sum of exports and imports of goods and services as a per cent of GDP).
A standard finding in the literature is that more open economies will have bigger
governments in order to protect their citizens against cyclical volatility in economic
activity. However, it could also be argued that in a globalising world small open
economies, due to their greater exposure to international competition, will be under
pressure to keep public expenditure low so as to secure competitive gains from a low tax
burden. Accordingly, the net effect on government size is ambiguous.
54