Tax Competition and the Enlargement of the EU A. Jevcak, Universty of Dortmund Introduction In the debate about potential benefits of differentiated tax systems in the current and future EU member countries arguments against tax competition rely predominantly on the models originated by Zodrow and Mieszkowski (1986) and Wilson (1986), and surveyed by Wilson (1999) which show that in the situation where governments can only attract more capital by lowering capital tax rates tax competition leads to under-provision of public goods. These models assume that countries levy taxes on capital in order to finance the provision of public goods for their citizens whose welfare increases in the level of public goods provided to them.1 It is not surprising that under such system a higher tax rate represents no benefits to capital which therefore moves to countries with lower tax rates decreasing welfare in the countries with higher tax rates. With respect to the design of the tax system Charles McLure (1986, pp. 342f) suggests that taxes should more closely reflect benefits received by taxpayers. Thus when providing public goods that benefit only citizens governments should rely more on taxes paid by individuals and less on capital taxation. This proposition is similar to Pascal Salin’s (1990, p. 201) view that the tax structure in a country should match the expenditure structure. Public goods provided to increase the welfare of citizens should therefore be paid for by taxation of citizens and supply of public goods to capital should be financed by capital taxation. Salin argues that different tax rates lead to capital flows only if there is a distortion between the expenditure and the revenue side, as decisions on capital location depend on the combination of taxes and benefits offered by the different countries. These arguments are in line with the perception usually shared in the U.S. which considers competition among states as a key element of federal system. According to Janeba and Schjelderup (2002, p. 3) this positive view of tax competition is perhaps a reflection of the influential article by Tiebout (1956) who argued that fiscal autonomy leads to an efficient outcome by allowing citizens to choose (“vote with their feet”) the combination of taxes and public goods best reflecting their preferences. In the light of these arguments it is not clear whether in the context of the enlarged European Union (EU) lower tax rates in the future member countries from central and eastern Europe (CEE) would by themselves lead to capital inflows into these countries as the level and quality of public goods in these countries is also lower. On the other hand, models of tax competition that assume that tax revenues are used for the provision of so-called industrial public goods/public inputs nevertheless suppose that governments are flexible in setting the levels of provided public goods and that they set these levels in a way to maximize the utility of citizens. 2 However, public inputs like infrastructure, 1 Zodrow, G.R. and P. Mieszkowski (1986), Wildasin, D. E. (1988), Wilson, J.D. (1991), Bucovetsky, S. (1991), Razin, A.and E. Sadka (1991), Persson, T. and G. Tabellini (1992); Zodrow and Mieszkowski (1986) consider both a model with consumable public goods and a model with industrial public goods. 2 Zodrow and Mieszkowski (1986), Bayindir-Upmann, Thorsten (1998), Fuest, Clemens (1995), Rauscher, Michael (1997), Rauscher, Michael (1998). education, national defence or effective police and legal system have to be built up over many years and thus their levels cannot be adjusted instantaneously. Moreover, although public inputs can be expected to benefit all primary (private) factors of production the models with public inputs usually assume that capital taxation is the only source of government revenue.3 Furthermore, these models consider a number of small identical jurisdictions which take the world rate of return on investment as given. In the context of the EU enlargement it might however be more interesting to analyse how the distribution of capital located in two unevenly-developed groups of countries could change after the creation of a single market. Hence, the issue of tax competition is reconsidered in this paper using a two-country framework where capital and labour tax revenues are used exclusively on the provision of public inputs while countries differ in the levels of public inputs accumulated over time. It is shown that when public inputs are financed both through capital and labour taxation then tax competition does not necessarily have to lead to under-provision of public goods. The analysis furthermore rises some doubts about the effects and the efficiency of possible capital tax harmonisation in the EU. A simple model There is an industrial public good which consists of a stock of public inputs accumulated over the years (GS) and of public inputs financed through primary input taxation (G T) with GT = TK + (1 )TL where T and (1 )T are tax rates on a unit of capital and labour employed in the given country, thus G = GS + TK + (1 )TL. (1) A two-country framework is modelled where a more developed country has a larger stock of government provided inputs than a less developed country, GS > GS*, where * denotes the less developed country. Capital stock is normalised to unity and perfectly mobile across the two countries (K = K, K* = 1 – K) while labour force although of equal size (L = L* = 1) in both countries is mobile only within a single country. Initially, more capital is located in the more developed country, that is K > 1/2. Public inputs can basically be divided into pure sometimes also called “factor-augmenting” public inputs and congestible public inputs. Pure public inputs are freely and fully available to all producers without being subject to congestion, either in the number of producers or in the quantities of input or output. Basic research, national defence, legal system or flood control could be considered as examples of pure public inputs. For this type of public input it seems appropriate to assume a production function with constant returns to scale in primary inputs (private) and increasing returns to scale in all inputs of production (private and public). On the other hand, the degree to which a congestible public input benefits each producer varies with the amount of utilisation by other producers. Infrastructure, education system or public administration are usually given as examples of congestible public inputs. In case of such public inputs a production function with diminishing returns to scale in primary factors and constant returns to scale in all factors of production should be more suitable. 3 Zodrow and Mieszkowski (1986), Bayindir-Upmann, Thorsten (1998) and Rauscher, Michael (1997). Fuest (1995) and Rauscher (1998) suppose different tax instruments but still do not introduce labour taxation. 2 As a result two versions of the model are considered in this paper with the specification of the production function determined by the type of public input. The case of pure public inputs There is a single aggregate good which can be either consumed or used as a production input. The aggregate good is produced according to a neoclassical production function Y(K,G,L) exhibiting constant returns to scale in primary inputs (K,L) and increasing returns to scale in all production inputs (K,G,L), with YK,YG,YL > 0, YKK,YGG,YLL < 0 and YKG,YKL,YGL > 0. As capital is perfectly mobile within the two countries after tax profit per unit of capital has to be the same in both countries, that is YK – T = YK* – T* (2) where subscripts indicate derivatives. Differentiating this capital market arbitrage condition with respect to the tax rate in the less developed country while substituting for G and G * from eq. (1) and solving for KT* gives KT* (1 K ) (1 ) YKG YKK YKG T YKK YKG T (3) In case of a production function of the form Y = KGL with + = 1, > and > the denominator is always negative.4 Hence, an increase in T* leads to a decrease in K when – YKG*((1 – K) + (1 )) < 05 which can be rewritten as 1 (1 K ) 1 G 1 (1 K ) (1 ) (1 ) YG (1 K ) (4) This result implies that in case of a relatively small G* or a small an increase in the tax rate in the less developed country can actually induce a capital inflow into this country. This is caused by the fact that the lower is G* the higher is the positive impact of additional public expenditures on capital productivity while the lower is the more are the cost of public inputs provision carried by labour. Tax competition thus does not necessarily have to lead to a race to the bottom as it is often argued. Furthermore, differentiating the right hand side of the condition (4) with respect to leads to 4 The denominator can be split into two parts which can be both shown to be negative as YKK YKG T ( 1) K ( 2) G K ( 1) G ( 1) T K ( 2) G ( 1) (( 1)G KT ) where G > KT while (1 – ) = > . Zodrow and Mieszkowski (1986) reach the same result in a model where =1. They arbitrarily assume that this condition is not fulfilled. 5 3 ( / )(1 K ) 1 G 1 ((1 K ) (1 )) ( / ) (1 K ) 1 G 1 ((1 K ) (1 )) Log[G ]. This term is positive as long as 1 Log[G ] 0 that is as long as G e S 1 . Hence, if a 1 lower bound on G S is assumed such that GS e which seems to be a reasonable assumption6 then an increase in the efficiency of the public sector () improves the chances that an increase in the capital tax rate will lead to a capital inflow into the given country. The model thus emphasises the importance of the size and the effectiveness of the public sector, which are often omitted from the tax competition debate. Welfare Maximisation Lets assume that the government of the less developed country is trying to maximise the level of capital located at home while taking the tax rate in the other country as given. The less developed countries might be interested in maximising capital inflows because they seek the technology transfer or a better access to foreign markets associated with foreign direct investment or just because they see it as the only source of job creation. The equation (3) implies that in the situation where (1 K ) (1 ) and thus 1 (1 ) YG YKG (1 K ) (5) the government of the less developed country does not have an incentive to adjust T* if at the same time this situation is a stable equilibrium. This is shown to be the case in Appendix 1. If + (1 – )/((1 – K)) > 1 that is if (1 K) or are sufficiently small then the equation (5) implies that YG* < 1. However, the optimal provision of public goods in autarky is given by YG = 1 as in this case domestic production maximising governments increase capital tax rates until the marginal product of industrial public good provision equals its marginal cost. Hence, it seems that in the model where public inputs are financed both by capital and labour taxation7 tax competition might actually lead to over-provision of public goods. If governments instead of capital inflows maximised the net national income without engaging in any wasteful public expenditures that is MaxT*(Y* – YK*((1 – K) – KL) – T*(1 – K) – (1 – )T*L where KL is the stock of capital owned by the residents of the less developed country then YK ( K T ) YG ( (1 K ) (1 ) L T K T ) (YKK ( K T ) YKG ( (1 K ) (1 ) L T K T ))((1 K ) K L ) YK ( K T ) (1 K ) T K T (1 ) L 0 1 0,0000454; even if = 0,5 then e 6 If = 0,1 then e 7 The equation (5) reduces to 1 Y G G => Y 1 0.13. 1 / 1 if it is assumed that public inputs are financed solely through capital taxation that is if G = GS + TK. 4 and thus K T ( T * YG T * (YKK YKG T * )((1 K ) K L )) YKG ( (1 K ) (1 ))((1 K ) K L ) (YG 1)( (1 K ) (1 )) 0 (6) This equation holds for YG*= 1, if (1 – K) = KL, hence, if such a situation was a stable equilibrium (which according to Appendix 2 cannot be ruled out) the net income maximising government would not have an incentive to change T* while government goods were provided in an optimal quantity. The case of congestible public inputs The aggregate good is now produced according to a neoclassical production function Y(K,G,L) exhibiting decreasing returns to scale in primary inputs (K,L) and constant returns to scale in all production inputs (K,G,L), with YK,YG,YL > 0, YKK,YGG,YLL < 0 and YKG,YKL,YGL > 0. The assumption of decreasing returns to scale in primary inputs implies that factor payments do not fully exhaust revenues and thus an economic rent is generated. The existence of a positive excess profit in a competitive market would however attract more private inputs until the excess profits were fully dissipated. It is therefore assumed that capital receives a share s of the economic rent generated by the public input. As capital is perfectly mobile within the two countries after tax profit per unit of capital has to be the same in both countries, that is YK + sYG(G/K) – T = YK* + sYG*G*/(1–K) – T* (7) where subscripts indicate derivatives while * denotes the less developed country. Differentiating this capital market arbitrage condition with respect to the tax rate in the less developed country now implies that 1 1 ) YKG ( (1 K ) (1 )) sYGG G ( ) 1 K 1 K KT* sY TK G s(YGG T YGK )G s(YGG T YGK )G sYG (G (1 K ) T ) YKK YKK YKG T YKG T G 2 K 1 K K (1 K ) 2 sYG ( Lets denote this expression as equation (8). The denominator of this expression can again be shown to be negative (Appendix A1) thus an increase in T* leads to a decrease in K as long as sYG ( 1 1 ) YKG ( (1 K ) (1 )) sYGG G ( ) 1 K 1 K This condition can be rewritten as (Appendix B1) 1 (1 K ) G ( 1) ( s 1 s 1 )( ) YG ( )( ) 1 K 1 K 5 (9) This result again shows that in case of a relatively small G* or an increase in the tax rate in the less developed country can actually induce a capital inflow into this country. The difference between equation (4) and (9) is the term s/ in the equation (9). As in the case of congestible public inputs capital receives the share s of the economic rent generated by the public input a higher s or imply a stronger positive impact of a tax increase on capital income. If s = 0, equation (9) reduces to (4) which is not surprising as in this case returns on capital are formally identical for both types of public inputs. Differentiating the right hand side of the condition (9) with respect to gives ( (1 K ) (1 )) 1 1 (1 K ) 1 G ( s )( (1 K ) (1 )) (1 K ) 1 G ( s )( (1 K ) (1 )) Log[G ] (1 K ) 1 G 1 s that is (1 K ) 1 G 1 s 1 ( (1 K ) (1 )) (1 K ) 1 G ( s )( (1 K ) (1 ))(1 Log[G ]) S S 1 The lower bound on G assumed in the previous case, G e , is thus again the sufficient condition to ensure that an increase in the productivity of the public sector () improves the chances that an increase in the capital tax rate will lead to a capital inflow into the given country. Welfare Maximisation Lets assume again that the government of the less developed country is trying to maximise the level of capital located at home while taking the tax rate in the other country as given. The equation (8) implies that in the situation where sYG ( 1 1 ) YKG ( (1 K ) (1 )) sYGG G ( ) 1 K 1 K and thus 1 (1 K ) G ( 1) ( s 1 s 1 )( ) YG ( )( ). 1 K 1 K (10) the government of the less developed country does not have an incentive to adjust T* if at the same time this situation is a stable equilibrium. This is shown to be the case in Appendix 3. If (/)( + (1 )/(1 K)) > 1 that is if is sufficiently small then the condition (10) implies that YG* < 1 for any value of s. Hence, when public inputs are financed both by capital and labour taxation8 then tax competition might lead to the over-provision of public goods both in the case of pure and in the case of congestible public inputs. 8 The condition (Y) reduces to 1 YG ( s ) YG 1 /( s ) 1 if it is assumed that public inputs are financed solely through capital taxation that is if G = GS + TK. 6 Empirical Evidence These conclusions seem not to contradict the empirical evidence which does not confirm any significant decreases in corporate income taxation despite deepening economic integration. Schulze and Ursprung (1999) surveyed a large number of empirical studies that examine the relationship between economic integration and corporate income taxation. They conclude that although a moderate downward trend in capital tax rates in the 1980s can be observed capital tax revenues have remained stable and governments continue to levy substantial capital tax rates (Schulze and Ursprung (1999), p. 321). Devereux, Griffith and Klemm (2002) analyse the development of taxes on corporate income in EU and G7 over the 1980s and 1990s. They show that while statutory tax rates fell over the 1980s and 1990s tax bases were on average broadened between the early 1980s and the end of the 1990s. As a result, the effective marginal tax rate has remained stable over the last two decades. Furthermore, although tax revenues on corporate income have declined as a proportion of total tax revenue since 1965 they have remained broadly stable as a proportion of GDP (Devereux, Griffith and Klemm (2002), p. 487). This finding is also supported by the Eurostat data on corporate income tax revenues in 14 EU countries between 1995 and 2001 (Table 1). Moreover, general government expenditure classified by the function of government (COFOG) can be divided into those that benefit both capital and labour and those that only benefit labour. The former group includes general public services, defence, public order and safety, economic affairs and environment protection while the latter consist of housing and community amenities, health, recreation, culture and religion, education and social protection. Table 2 shows that except for the housing and community amenities the relative average shares of labour benefiting categories of government expenditures in the EU increased between 1995 and 2001 (the share of the EU government expenditures on housing and community amenities would however remain unchanged if Netherlands which had an unusually high expenditure in this category in 1995 was not considered). The EU governments thus do not seem to experience significant problems neither with capital tax revenues nor with provision of labour benefiting public goods. Observed decreases in corporate tax rates might furthermore be attributed to ceasing of the inefficient industrial public good provision while an increased reliance on personal income taxation might be explained as an effort not to finance public goods benefiting citizens through corporate income taxation. In order to confirm this interpretation a precise empirical analysis of public expenditures is necessary but such study is difficult to perform.9 Capital Tax Harmonisation 9 Hines (1999, p. 309) states that one of the reasons why no significant link between taxes and FDI might be found is that governments imposing high tax rates may indirectly compensate firms with difficult-to-measure investment incentives such as worker training and infrastructure. 7 Does this model provide any support for capital tax harmonisation in the EU? No. Firstly, the optimal autarky condition for industrial public good provision YG = 1 is in this model associated with a different optimal tax rate (TO) in each country as YG K 1 (GS TO K )1 implies that 1 TO 1 ( K ) GS K A harmonised tax rate would therefore by itself not result in an optimal provision of public goods in each country if autarky situation is taken as a reference point. And secondly, YKG > 0 implies that an increase in GS makes the given country a more attractive location for investment. Hence, if capital tax rates were the same in both countries, more capital would be located in the country which offered more industrial public goods. Therefore, if capital tax rates in the EU were equalised before the stock of industrial public goods reached the same level in all member states the less developed countries would not end up with the same amount of capital as the more developed ones. As a result, capital tax revenues in the less developed countries would be lower. The EU would then have to rely on fiscal transfers from the more developed countries if it wanted to equalise the stocks of industrial public goods and thus capital in all member states. Conclusion The simple model presented in this paper shows that when industrial public goods are financed both through capital and labour taxation then tax competition does not necessarily have to lead to under-provision of public goods. In their effort to attract more capital countries might actually over-provide industrial public goods. The model also implies that when tax rates in two regions are harmonised before the stock of accumulated industrial public goods is the same in both regions then it is impossible for a less developed region to build up its stock of public goods without fiscal transfers from the more developed region. (Although a closed form solution of a Nash game between the two welfare maximising governments would be very nice I do not think this is possible or at least I do not know how to reach such solution. I will try to come up with some reasonable limitations that would allow for a further analyses of the model. Some numerical simulations could be performed but I do not think they would be very persuasive.) jevcak@gmx.de 8 References Bayindir-Upmann, Thorsten (1998), ‘Two Games of Interjurisdictional Competition When Local Governments Provide Industrial Public Goods’, International Tax and Public Finance, Vol. 5, pp. 471-487. Bucovetsky, S. (1991), ‘Asymmetric Tax Competition’, Journal of Urban Economics, Vol. 30(2), pp. 167-81. Fuest, Clemens (1995), ‘Interjurisdictional Competition and Public Expenditure: Is Tax Coordination Counterproductive?’, Finanzarchive NF Bd. 52. Devereux, Michael P., Rachel Griffith and Alexander Klemm (2002), ‘Corporate Income Tax Reforms and International Tax Competition’, Economic Policy - CEPR, CES, MSH, pp. 450495. Persson, T. and G. Tabellini (1992), ‘The Politics of 1992: Fiscal Policy and European Integration’, Review of Economic Studies, Vol. 59(4), p. 689, 13p. Rauscher, Michael (1997), ‘Interjurisdictional Competition and the Efficiency of the Public Sector: The Triumph of the Market over the State?’, CEPR Discussion Paper No. 1624, London: The Centre for Economic Policy Research. Rauscher, Michael (1998), ‘Leviathan and Competition among Jurisdictions: The Case of Benefit Taxation’, Journal of Urban Economics, Vol. 44, pp.59 - 67. Razin, A.and E. Sadka (1991), ‘International Tax Competition and Gains from Tax Harmonization’, Economic Letters, 37(1). Schulze, Günter G. and Heinrich W. Ursprung (1999), ‘Globalisation of the Economy and the Nation State’, World Economy, Vol. 22(3), pp. 295-352. Wildasin, D. E. (1988), ‘Nash Equilibria in Models of Fiscal Competition’, Journal of Public Economics, Vol. 35(2), pp. 229-40. Wilson, John D (1986), ‘A Theory of Interregional Tax Competition’, Journal of Urban Economics, Vol. 19, pp. 296-315. Wilson, John D. (1991), ‘Tax Competition with Interregional Differences in Factor Endowments’, Regional Science and Urban Economics, Vol. 21, pp. 423-52. Wilson, John D. (1999), ‘Theories of Tax Competition’, National Tax Journal, Vol. 52, Issue 2, p. 269, 36p. Zodrow, G.R. and P. Mieszkowski (1986), ‘Pigou, Tiebout, Property Taxation, and the Underprovision of Local Public Goods’, Journal of Urban Economics, Vol. 19(3), pp. 356-70. McLure, Charles E. (1986), ‘Tax Competition: Is What’s Good for the Private Goose also Good for the Public Gander?’, National Tax Journal, Vol. 39, pp.341 - 348. Salin, Pascal (1990), ‘Comment on Vito Tanzi and A. Lans Bovenberg, “Is There a Need for Harmonizing Capital Income Taxes within EC Countries?”’, pp.198 – 205 in: Siebert, Horst (Ed.) (1990), Reforming Capital Income Taxation, Tübingen-Mohr. Janeba, Eckhard and Guttorm Schjelderup (2002), ‘Why Europe Should Love TaxCompetition and the US Even More So’, NBER Working Paper 9334 9 Appendix A1 1 1 ) YKG ( (1 K ) (1 )) sYGG G ( ) 1 K 1 K sY TK G s(YGG T YGK )G s(YGG T YGK )G sYG (G (1 K ) T ) YKG T YKG T G 2 K 1 K K (1 K ) 2 sYG ( KT* YKK YKK YKK YKK YKG T YKG T sYG TK G s(YGG T YGK )G s(YGG T YGK )G sYG (G (1 K ) T ) 0 K 1 K K2 (1 K ) 2 The denominator can be split into two equivalent parts which can be shown to be negative. Y KK s(YGG T YGK )G sYG (G (1 K ) T ) Y T 0 1 K (1 K ) 2 KG ( 1)(1 K ) ( 2 ) G (1 K ) ( 1) G ( 1) T s ((1 K ) ( 1)G ( 1) T (1 K ) ( 1) G ) s(1 K ) G ( 1) (G (1 K ) T ) 0 1 K (1 K ) 2 (1 K ) ( 2) G ( 1) ( 1)G (1 K ) T ( s( 1) ) sG s (G (1 K ) T ) 0 (1 K ) ( 2) G ( 1) ( 1)G (1 K ) T ( s ) s( 1) G 0 (1 K ) ( 2) G ( 1) (( 1)G (1 K ) T ) s ((1 K ) T ( 1)G ) 0 Since per definition G* > (1 K)T* while (1 – ) > the expression in brackets and thus the whole term is negative. The same procedure can also be applied to the second part of the denominator. Appendix B1 sYG ( 1 1 ) YKG ( (1 K ) (1 )) sYGG G ( ) 1 K 1 K ( sYG YKG (1 K ) sYGG G )( 1 ) 1 K 10 ( s(1 K ) G ( 1) (1 K ) G ( 1) s(1 K ) ( 1)G ( 1) )( 1 (1 K ) G ( 1) ( 1 ) 1 K s 1 s 1 )( ) YG ( )( ) 1 K 1 K Appendix 1 Differentiating the Equation (3) with respect to T* at the point where KT* = 0 gives (1 K ) (1 ) YKGG 0 YKK YKG T YKK YKG T 2 K T *T * As both the numerator and the denominator are negative, this expression is always positive. As a result, K(T*) is convex at KT* = 0 and thus the less developed country does not have an incentive to change T* at this point. Appendix 2 Differentiating the Equation (6) with respect to T* at the point where YG*= 1 and (1 – K) = KL gives K T (YKK YKG T * )( K T ) YKG ( (1 K ) (1 )( K T ) (YGk ( K T ) YGG ( (1 K ) (1 ) T * K T ))( (1 K ) (1 )) that is KT (YKK YKG T * )(KT ) YGG ( (1 K ) (1 ) T * KT ))( (1 K ) (1 )) While the first term is always positive (see footnote 7) the second term is negative if K T* is negative which as implied by equation (5) can not be ruled out for YG* = 1. As the relative size of the two terms is unclear it is not possible to tell whether the whole expression is positive or not. Appendix 3 Differentiating the Equation (8) with respect to T* at the point where KT* = 0 gives 1 1 ) YKGG ( (1 K ) (1 )) 2 sYGGG G ( (1 K ) (1 ))( ) 1 K 1 K sY TK G s(YGG T YGK )G s(YGG T YGK )G sYG (G (1 K ) T ) YKK YKK YKG T YKG T G 2 K K 1 K (1 K )2 2sYGG ( (1 K ) (1 ))( KT *T Using 11 G YGGG ( 1)( 2)G ( 2) K ( 2)YGG leads to 1 ) YKGG ( (1 K ) (1 )) 2 1 K 0 sYG TK G s(YGG T YGK )G s(YGG T YGK )G sYG (G (1 K ) T ) YKK YKK YKG T YKG T K2 K 1 K (1 K )2 sYGG ( (1 K ) (1 ))( KT *T As both the numerator and the denominator are negative, this expression is always positive. As a result, K(T*) is convex at KT* = 0 and thus the less developed country does not have an incentive to change T* at this point. 12 Table 1: Taxes on the income or profits of corporations as percentage of GDP Belgium Denmark Germany Greece Spain France Ireland Italy Luxembourg Netherlands Austria Portugal Finland Sweden United Kingdom 2001 3.2 3.1 : 3.2 3.0 3.1 3.6 2.9 7.5 4.4 3.3 3.6 4.3 3.7 3.3 2000 3.3 2.4 : 4.4 3.2 2.8 3.8 2.4 7.2 4.4 2.2 4.1 6.0 2.9 3.3 1999 3.3 3.0 : 3.3 3.0 2.7 3.8 2.8 7.0 4.6 2.0 3.8 4.4 2.9 3.3 1998 3.4 2.8 : 3.1 2.6 2.3 3.4 2.5 7.8 4.5 2.3 3.3 4.3 3.0 3.6 1997 2.9 2.6 : 2.6 2.8 2.3 3.2 4.2 7.9 4.6 2.2 3.3 3.5 2.7 3.4 1996 2.7 2.3 : 2.3 2.1 2.0 3.1 3.8 7.7 4.1 2.2 2.9 2.8 2.9 2.7 1995 2.4 2.0 : 2.6 1.9 1.8 2.8 3.4 7.5 3.3 1.7 2.5 2.3 1.9 2.4 Source: TAX_AGR Main national accounts tax aggregates, EUROSTAT, LUXEMBOURG 13 14 Table 2: COFOG Percentage of Total General Government Expenditure General public services 1995 Be Belgium 22,7 Dk Denmark 7,3 De Germany 11,9 gr Greece 43,7 es Spain .. fr France 11,4 ie Ireland 8,4 it Italy 26,4 lu Luxembourg 10,1 nl Netherlands 17,7 at Austria 16,4 pt Portugal 19,3 fi Finland 11,6 se Sweden 17,6 Uk UK 13,1 2001 20,0 8,1 13,1 33,7 .. 12,2 10,4 19,7 11,8 17,6 16,4 14,3 13,0 14,9 10,9 Defence 1995 2,8 3,0 2,5 5,7 .. 5,4 2,7 2,2 1,3 3,4 1,7 4,9 3,4 3,8 7,1 EU-14 Average 17,0 15,4 3,6 2001 2,4 3,1 2,5 6,7 .. 4,6 2,1 2,3 0,8 3,4 1,7 3,7 3,0 3,8 6,6 3,3 Housing Public Recreation, Economic Environment and order and Health culture and Education affairs protection community safety religion amenities 1995 2001 1995 2001 1995 2001 1995 2001 1995 2001 1995 2001 1995 2001 2,8 3,2 8,9 8,9 1,3 1,4 0,6 0,8 11,7 13,4 1,7 2,0 12,1 12,6 1,7 1,8 7,6 6,5 0,0 0,0 1,5 1,6 8,5 9,8 2,8 3,1 12,8 15,0 3,0 3,3 20,3 8,9 1,8 1,2 1,6 2,3 11,4 13,1 1,4 1,5 8,0 8,7 2,2 2,3 3,5 6,2 1,0 0,8 0,8 0,8 4,7 5,4 0,4 0,6 6,3 7,9 .. .. .. .. .. .. .. .. .. .. .. .. .. .. 1,8 1,9 11,6 9,9 2,0 2,5 1,6 1,9 14,3 15,0 1,5 1,5 11,4 11,4 4,1 4,5 26,7 21,4 0,0 0,0 4,3 6,8 14,9 18,8 1,0 1,8 12,3 12,8 3,9 3,9 8,6 8,3 1,3 1,7 1,7 1,9 10,3 13,0 1,5 1,9 9,2 10,6 1,8 2,3 11,6 7,2 3,3 3,3 2,4 2,1 12,3 12,6 3,7 4,4 11,0 12,1 2,5 3,2 8,7 12,0 1,4 1,5 12,1 3,2 6,9 8,8 1,6 2,4 9,0 10,3 2,6 2,7 10,1 10,5 2,1 0,8 1,7 1,9 13,8 11,1 1,6 1,9 11,0 11,1 4,7 4,1 12,4 13,2 0,9 1,5 1,6 1,9 11,8 14,7 2,4 2,6 14,4 14,9 2,5 2,8 11,8 9,3 0,5 0,6 1,7 1,2 10,4 12,2 2,2 2,4 12,3 13,0 2,2 2,4 9,0 7,7 0,3 0,5 4,3 1,7 9,4 11,5 2,8 1,9 10,5 13,5 4,8 4,8 7,4 5,9 0,7 1,3 2,5 1,0 12,9 15,5 1,4 1,3 10,3 11,7 2,9 3,1 11,3 9,7 1,2 1,2 2,7 2,1 11,0 12,5 1,9 Source: PUBL_EXP General government expenditure by COFOG function and by type, EUROSTAT, LUXEMBOURG Social protection 1995 35,2 44,4 38,0 32,0 .. 39,0 25,3 35,0 42,2 36,7 38,6 27,8 43,8 40,3 39,8 2001 35,2 43,8 45,2 35,8 .. 38,9 21,4 36,6 43,6 37,6 41,9 29,0 42,1 41,8 40,7 2,1 10,8 11,8 37,0 38,1