The Treasury

Global Navigation

Personal tools


Fiscal Policy, Growth and Convergence in Europe - WP 03/14

3  Estimating Fiscal Policy Impacts on Growth in Europe

Using the framework discussed in section II, Kneller et al (1999) and Bleaney et al. (BGK, 2001) estimated the growth impacts of fiscal variables on long-run growth in OECD countries. In this Section we apply the BGK (2001) results to EU-specific data. Firstly, we consider the effect on average growth rates of the changes made to the fiscal budget in various European countries over the 1990s.[7]

The estimates made here rely on correctly identifying structural from cyclical changes to policy. To try to compare similar points in the business cycle we use data from the end of the 1980’s and the latter part of the 1990s and average across 3-year periods to minimise any remaining business cycle effects. To maintain consistency with BGK the work here is based on IMF data at the consolidated central government level. Data limitations reduce the sample to 11 European countries. To these we add the US - the small size of its public sector compared to European countries makes it a useful comparison. The eleven European countries comprise nine members of the European Union (of which six are Euro-members), plus Norway and Switzerland. The BGK (2001) estimates are based on a sample of 17 OECD countries over 1970-94, using a dynamic fixed effects panel with annual data. A long lag structure (8 years) was included to separate the short-run and long-run effects of policy. The fiscal data were pre-classified to match the types of fiscal variable in Table 1. These are given in Table 2.

As noted above, growth theory predicts that unproductive expenditure and non-distortionary taxes have no effect on growth and can therefore be removed from the regression equation. Empirical testing contained in BGK (2001) supports this claim. The estimated parameters of the remaining five fiscal variables (along with their standard errors) are listed in Table 3 below. These results suggest that increasing tax revenues from distortionary taxes by 1 per cent of GDP reduces the average growth of the economy by 0.411 percentage points, whereas increasing productive expenditures by the same amount increases growth by 0.387 percentage points.[8]

Table 2 - Classifying Taxes and Expenditures



Unproductive ExpenditureOther Expenditure
Education Social security & welfare Other expenditures
Health Recreation  
Law & order Economic services  
General public services   
Transport & communication   

Table 2 - Classifying Taxes and Expenditures - continued
Distortionary TaxationNon-distortionary TaxationOther Revenues
Income & profit taxes Domestic goods & services taxes International trade taxes
Social security taxes  Other tax revenues
Payroll & manpower taxes  Non-tax revenues

Note: The budget surplus/deficit is calculated as total revenues less total expenditures.

Table 3 – Parameter Estimates from Bleaney et al. (2001)
Fiscal variable

Parameter estimate

(standard error)

Budget surplus (surp/def)


Distortionary taxation (rdis)


Productive expenditure (eprd)


Other expenditures (eoth)


Other revenues (roth)



3.1  EU Fiscal Policy in the 1990s

Table 4 provides some evidence on taxes and expenditures across the sample countries. During the period 1995-97 the government that appropriated on average the greatest proportion of GDP was the Netherlands, although several of the other European countries were not far behind. According to Table 4 the public sector, measured by central government total expenditure, was greater than 40% of GDP on average during the period 1995-97 in Austria, Denmark, France, Norway, Netherlands, Sweden and the UK. Perhaps unsurprisingly the US has the smallest government sector of the countries considered here.

The greater portion of tax revenues in the sample are collected from distortionary taxes: an average of 20.7% of GDP against an average for non-distortionary taxes of 9.4%. This varies between over 30% of GDP in the Netherlands to 13% in Finland. Norway collects the greatest proportion of its GDP from non-distortionary taxes (15.4%), and the US the least (0.7%). The figure for the US is low relative to the other countries in the sample partly because most indirect taxes are issued at the local rather than the national level.

Unlike the revenue side of the budget, expenditures are more evenly split between productive and unproductive forms. The average for all 12 countries is 13.2% for productive expenditure and 18% for unproductive expenditure. The most generous welfare systems are in the Scandinavian and Western European democracies such as Sweden and France. The least generous are in Spain and the US, again in line with expectations. The greatest provision of productive expenditure is made in the Netherlands while the lowest is again in Spain.

According to Table 4 all of the countries in the sample, except Norway, had on average a budget deficit during the mid-to-late 1990s. Given the timing of the data used to calculate Table 4 it is perhaps unsurprising that the annual deficit in the EU countries is around the EMU-convergence limit of 3% of GDP.

In order to estimate how changes made to fiscal policy over the 1990’s affected the average growth rate of the European economies we must separate structural from cyclical changes to fiscal policy. Table 5 does this for each of the fiscal categories by subtracting the average for 1987-89 from the 1995-97 average. The period 1987-89 was chosen as it represents a broadly similar point in the business cycle and the data were also averaged across 3-year periods to minimise any remaining business cycle influences.[9]

Table 4 - Government Budget Data (% GDP), average for 1995-97
 rdis rndis


of which:



enprd surp/ def texp trev
St. dev.

Note: All figures are expressed as ratios of GDP. The ‘other revenues’ and ‘other expenditure’ categories are omitted from the Table to conserve space. rdis = distortionary tax revenues; rndis = non-distortionary tax revenues; eprd = productive expenditure; enprd = unproductive expenditure; educ = education expenditure; ehlth = health expenditure; surp/def budget surplus/deficit; texp = total expenditure; trev = total revenues.

Table 4 also makes clear that the mix of fiscal revenues and expenditures differs markedly between countries, while Table 5 shows that no single factor dominated changes in the fiscal budget over the 1990’s. Of the few likely common influences the political pressure to meet the convergence criteria for EMU entry is an obvious one. This shows up only weakly in the data however. The annual deficit fell as a percentage of GDP in Germany, Austria and the Netherlands, while it rose in France, Finland and Spain. In these latter countries the deficit was lower in 1997 than in 1995 suggesting that the process of restructuring was not complete in the data available here. Of the three EU countries in the sample that chose not to join EMU, the UK, Sweden and Denmark, all saw a rise in the annual deficit as a percentage of GDP compared to the late 1980’s. It would seem that any political pressure was less of a binding constraint in these countries.

Table 5 – Changes to the Government Budget
(% GDP; 1987-9 to 1995-7)
 rdisrndiseprdenprdsurp/ def

Note: rdis = distortionary tax revenues; rndis = non-distortionary tax revenues; eprd = productive expenditure; enprd = unproductive expenditure; educ = education expenditure; ehlth = health expenditure; surp/def budget surplus/deficit.

Despite the rather complex set of changes to the fiscal budget described in Table 5 the public sector (measured by total expenditure) increased by more than 1 percentage point of GDP in only four of the twelve countries: Sweden (1.9), the UK (2), Finland (3.4) and Spain (6.2). It fell in Denmark (-0.7), Norway (-0.9) and the US (-0.2), although none by more than 1 per cent of GDP. Given the decline in the deficit noted above it would appear from this that most European countries chose to reduce the deficit by increasing revenues rather than decreasing expenditures.

In some cases expenditures on productive or unproductive goods and services were increased, perhaps to offset any negative political consequences of raising taxes. In Germany tax revenues were increased along with unproductive expenditures, while in Austria it was productive expenditures which were raised. The Netherlands is something of an exception: expenditures and distortionary tax revenues both fell (though again this may reflect a tendency to use some policy changes to offset the negative political consequences of others). The net effect of these changes was to increase the size of the public sector by 0.4 percentage points in Austria, 0.7 percentage points in Germany and 0.1 percentage points in the Netherlands. A similar lack of consistency in terms of policy rules appears to have occurred in the EMU-member countries in which the deficit rose.[10]


  • [7]To simulate tax, expenditure and deficit changes simultaneously and at a reasonable level of detail from the available literature we are limited to using the results in Kneller, et al. (1999) and BGK (2000). We consider the latter to be more reliable.
  • [8]Note that the estimated effect of the budget surplus on long-run growth is somewhat smaller (at 0.105) and, by assumption this figure applies regardless of the method of financing budget deficits (e.g. money creation, bond sales). It might be expected that money-financed deficits would have especially harmful effects on growth (though not necessarily over the long-run). However, macroeconomic policy in OECD countries in the 1990s has generally avoided significant money-financing of budget deficits. Note also that omitting ‘neutral’ categories (non-distortionary taxes, unproductive expenditures) from Table 3 implies that these do not affect long-run growth but they may still affect the long-run level of GDP per capita.
  • [9]These calculated changes to policy are amended slightly to ensure adding-up across the budget constraint. An identical data span was not available for all countries and so we compare the periods 1989-91 and 1993-95 for Denmark, and the latest suitable data for France, Germany and Switzerland is 1991-93. The estimated changes to fiscal policy in these countries are therefore more likely to contain information about cyclical adjustments.
  • [10]In France the increase in the deficit was matched by a decrease in revenues from non-distortionary taxation but expenditure on productive goods and services were increased. In Finland the increased deficit appears to have largely been used to fund increases in unproductive expenditure, while in Spain there was a large increase in other expenditure and some decrease in productive expenditures.
Page top