4 Convergence and Divergence in EU Fiscal Policy (continued)

This raises the question of whether a European sub-sample shares common steady-state values? If EU countries, as is often suggested, share similar fiscal constraints and have been following harmonisation processes over this period, it might be expected that their long-run expenditures, taxes or surpluses (as shares of GDP) would be similar. On the other hand, different characteristics across EU countries (such as different population age structures, differing expenditure preferences) may perpetuate differing fiscal values in the long-run.

To examine this, we proceed as follows. First, using pair-wise comparisons, we consider all pairs of countries for which the hypothesis, α_k = α_{j(k ≠ j) can be accepted. Following a similar procedure for groups of three, four etc countries sharing the same α, we identify the maximum number of countries belonging to such a group. For the three fiscal variables that appear in our growth regressions, this leads to the classification of countries given in Table 11 below. The F-statistics at the foot of each column test the hypothesis that countries belong to the groups to which they have been allocated.}

Perhaps the most interesting case concerns the budget surplus, where 7 of the 10 EU countries appear to share the same steady-state values (fixed effects), with only Denmark, Germany and Luxembourg excluded (the first two of which share similar values). For rdis and eprd, shared αs are less common with, at most, a group of 4 (rdis) or two groups of three (eprd) evident. For countries labelled “different α”, we can accept the hypothesis that their estimated fixed effects are significantly different from all other EU10 countries. That is, each country is estimated to converge on its own steady-state value of rdis, eprd or surp/def.

Two points emerge from these results. Firstly, it is not surprising that some EU countries choose similar long-run ratios to GDP for their productive public expenditures or distortionary taxes, whilst some choose quite different values. Secondly, evidence of β-convergence (but the general absence of σ-convergence) suggests that observed conditional convergence primarily reflects the tendency for countries to revert to their long-term trend, rather than any tendency for countries to approach a common steady-state over the period.

Since the data available here terminates in 1995, it may be that much of the pressure on EU governments towards similar tax/expenditure patterns post-dates the current evidence. Nevertheless, for the period we can study, the results in this section suggest three things. (1) There is very little evidence for σ-convergence - relevant fiscal variables are not, in general, becoming more similar across Europe. (2) There is only limited evidence of EU countries sharing the same steady state values of the three growth-affecting fiscal variables – rdis, eprd, and surp/def.[20] (3) Evidence of β-convergence implies that country fiscal variables tend to revert toward their steady-state paths.