Proceedings from the ECFIN Workshop "The budgetary implications of structural reforms" - Brussels, 2 December 2005

34. The regression results for the controls, without and with country-fixed effects, are shown in the
first two columns of Table 4. The inclusion of country-fixed effects indeed proves to be necessary; without
them the significance levels are unsatisfactory except for the debt ratio and the lagged dependent variable.
In the equation with country-fixed effects, the regression coefficients for the dependency ratio, per capita
income and public indebtedness have the expected sign and are all significant. Trade-openness has a
negative sign, and is also significant.

35. The long-run coefficients, which are the more interesting ones to look at, can be computed as
γ_k∕(1-λ). Based on this calculation, the long-run semi-elasticity on the dependency ratio turns out to be about
¼, which suggests that if in any country old-age population growth outpaces total population growth by
1 percentage point the primary expenditure ratio will increase by ¼ percentage point. Similarly, an increase
in per capita income by 10% would raise the primary expenditure ratio by 1½ percentage point and if
foreign trade (imports + exports) outpaces GDP by 10 percentage points for any given period or country,
the primary expenditure ratio would fall by 1½ percentage point. Finally, an increase in the debt ratio to
GDP by 10 percentage points would depress the primary expenditure ratio by approximately 1¼ percentage
point. These findings are broadly in line with those reported by Martinez-Mongay (2002), except for the
trade-openness indicator for which he finds the opposite sign. This may reflect the difference in estimation
period (1960-99 in his case versus 1985-2003 in ours), since it may be that over time competitiveness
considerations have gained importance as globalisation progressed. It is also in line with theoretical
findings by Buti and van den Noord (2005) that there may be incentives for small open economies to aim
for smaller governments as this would enhance the working of automatic stabilisers.

Table 4. Estimation results

Note: *, ** and *** denote significance at 90, 95 and 99%, respectively.

36. Another issue is whether the enforcement of the Stability and Growth Pact and the run-up towards
the adoption of the single currency subsequent to the Maastricht Treaty has had a significant additional
impact on primary expenditure in Europe. To test this, the same regression was run including dummy

56

<<   <   59   60   61   62   63   64   65   >   >>