Social protection in Europe:
do we need more co-ordination?
Kees Goudswaard and Bart van Riel*
This article addresses the question of whether we need to accelerate social convergence in the enlarged EU by setting common standards for social benefits. We conclude that the case for this type of harmonisa tion is not strong. Analysing several indicators, we show that there are no clear signs of a social race to the bottom yet. Nor do we expect one as a consequence of the enlargement. Once economic convergence ma terialises, the new member states should also be able to converge to higher protection levels. We do how ever acknowledge the importance of the social dimension of the integration process and argue that the current method of open co-ordination in the social field should be strengthened to face the common chal lenge of reforming the social system. It is not more but more effective co-ordination that is needed. Keywords: social protection, co-ordination, policy competition, social convergence
Introduction
The various national social protection systems are deeply embedded in the member states of the European Union (EU). The principles and organisation of social security are national re sponsibilities, but the social dimension has a place of its own in the European integration process. Social convergence in the EU has long been an important goal. A new and important step was taken with the adoption of the strate gic Lisbon Agenda in 2000, explicitly coupling the economic objective of becoming the most competitive and dynamic economy with the social objective of greater social cohesion. This is a major challenge, and even more so in the enlarged EU, and one that requires the moder nisation of the European welfare states. Once the Constitution has been accepted, social pro tection and social cohesion will also become constitutional goals of the EU.
The open method of co-ordination is ap
plied to social policy, which means the member states define and evaluate their common objec tives. The objectives pertain to employment, social inclusion and pension reform. This method is based on benchmarking and peer pressure, but does not provide any possibilities for sanctions. The EU does not prescribe how to achieve the objectives. This is still a matter of national sovereignty under the principle of subsidiarity, which means decision making takes place at the lowest level of government that is appropriate for the particular issue. But what level of government is appropriate for so cial policy? In other words, do we need more social protection policy co-ordination? This is the central question in this article, especially in the context of the enlargement. Should the enlarged EU try to accelerate convergence by setting common standards? We approach this question from a theoretical as well as an em pirical point of view.
Theoretically, there can be grounds for co ordination or even harmonisation if there are
* K.P. Goudswaard is professor of economics and social security at Leiden University (e-mail: k.p.goudswaard@ law.leidenuniv.nl), B. van Riel is senior economist at the Social and Economic Council (SER) and a Fellow at AIAS
(e-mail: riel@gw.ser.nl). Both contributed to the recent SER report With Europe More Growth, parts of which
Social protection in Europe
decentralisation failures such as international spillovers or externalities and economies of scale. These failures should outweigh the bene fits of decentralisation such as diversity in pre ferences and accountability. As far as social policy is concerned, the main argument in fa vour of more co-ordination is the existence of externalities. Labour and capital mobility in a fully integrated economic space may lead to forms of competition that could result in lower protection levels, i.e. the well-known race-to- the-bottom argument. Due to the dangers of social dumping and benefit tourism, the intro duction of European minimum social stan dards such as minimum benefit levels has of ten been proposed (Hutsebaut 2003). It has however also been argued that there are no clear signs of a race to the bottom actually tak ing place. In addition, the costs of premature harmonisation can be extremely high (Sinn and Ochel 2003).
After briefly discussing the history of Social Europe in the next section, we examine the theoretical discussion on social policy co-ordi nation and the present empirical evidence on the race-to-the-bottom argument. In the past few decades, social protection levels do in fact appear to have converged to a common higher level. We then discuss the impact of the enlar gement on social policies. Do lower protection levels give the new member states a competi tive advantage? Does this threaten the social systems in the old member states? In the last section, we draw some conclusions. We argue that the open method of co-ordination pro vides a good balance between subsidiarity and the supposed need for a stronger European so cial policy. But if it is to be effective, this meth od and its implementation need to be strength ened.
Social Europe
From Rome to Lisbon
The founding fathers of the EEC already ac knowledged the importance of the social di mension of the European integration process. However, since they expected social progress to result from economic integration, the Eu ropean Treaty of 1957 only provides a legal ba sis for harmonising social policies on the free movement of labour. Regulations on the social
protection of migrant workers were accepted as early as 1957. The most important regulation in this field, 1408/71 (recently replaced by 883/2004), on the co-ordination of social se curity systems pertaining to migrant workers, was passed after the free movement of workers was fully achieved at the end of the 1960s. As Leibfried and Pierson (1996) note, in conjunc tion with the accompanying case law of the Court, this regulation has restricted the sover eignty of member states in social policy.1 The emphasis is nonetheless clearly on the co-ordi nation rather than the harmonisation of social policy. The reconsideration (4) of regulation 8 8 3 /2 0 0 4 now explicitly cites the need to re spect the specific features of national social se curity legislation and only draw up a system of co-ordination.
EU directives were issued in the 1970s on the equal treatment of men and women and on safety and health regulations. But with the ex ception of issues involving the social protec tion of migrant workers, the social protection systems remained matters of national sover eignty. This has not changed. The Amsterdam Treaty of 1997 and the addition of the Social Protocol to the basic Treaty were steps forward in the social field in general, though they are not a basis for EU involvement with social pro tection levels in the member states.
The member states have nonetheless ex pressed a certain degree of commitment in terms of social protection. This commitment is embodied in two recommendations accepted by the European Council in 1992. The first one, dated June 1992, dealt with common criteria pertaining to sufficient resources and social as sistance in the social protection systems (92/ 441/EEC). The second one, dated July 1992, ex plicitly addressed the 'convergence of social protection objectives and policies' (9 2/442/ EEC). The idea was that convergence was to guarantee and stimulate the development of social protection in the context of the comple tion of the internal market. In addition, since the member states were faced with common problems such as ageing populations, unem ployment, changing family structures and pov erty, the common objectives were to act as pointers in the modification of the social pro tection systems addressing these problems. The recommendation also stipulates broadly defined goals, but 'without prejudice to the
powers of the member states to establish the principles and organisation of their systems'. Lastly, it recommends monitoring the progress in relation to the convergence of social protec tion aims and policies across the Union.
The desirability of member state policy con vergence has been reconfirmed in several Eur opean Commission reports such as the 1994 White Paper on European Social Policy and re ports on Social Protection in Europe. The 1998 Employment Guidelines drawn up as a result of the Jobs Summit in Luxembourg at the end of 1997 can partly be viewed as an implementa tion of the convergence strategy. One main plan proposed in the guidelines was to improve the employability of the unemployed. It re flected a shift in the systems of social protec tion towards a more active policy designed to get people working rather than merely trans ferring income to the unemployed. Though crucial in modernising social protection, this shift did not affect the benefit levels, coverage or eligibility criteria, i.e. the scope and gener osity of the national systems.
A new and important step was taken at the European Council in Lisbon in 2000. The stra tegic EU goal set for the decade ending in 2010 was to become the most competitive and dy namic knowledge-based economy with sus tainable economic growth and greater social cohesion. The economic and social agendas were explicitly coupled, and to achieve the aims, the social model was to be modernised. To ensure long-term sustainability of the so cial security systems in the light of the ageing population, participation rates had to be in creased.
The open method of co-ordination
The Treaty of Nice took the social agenda for ward. It was agreed to advance social policy on the basis of the open method of co-ordination (OMC). The OMC was first used with respect to employment policies in the EU, but as re gards social policy, the foundation was laid in the recommendation of July 1992. The method recognises that social policy remains the re sponsibility of the member states under the principle of subsidiarity. It implies that the member states define and evaluate common objectives and learn from each other what the best way is to reach the objectives. Best prac tices are disseminated and benchmarking is
used. Co-ordination is based on evaluation and peer pressure, but does not provide the op tion of sanctions. The decision was made in Nice that the member states were to imple ment action plans for combating poverty and social exclusion and define common objectives on social indicators. The indicators pertain to financial poverty, income inequality, long term unemployment, regional variations in employment rates, life expectancy and poor health. Some authors feel the common indica tors and national action plans for social inclu sion represent significant progress towards so cial integration (Atkinson 2002). Others ques tion this form of co-ordination. De Mooij and Tang (2003) argue that the absence of binding agreements may render social policy co-ordi nation ineffective, in which case the fear of so cial dumping may again lead to calls for har monisation.
Three forms of social policy co-ordination in
the EU
The developments cited above have resulted in three forms of social policy co-ordination: - Co-ordination of social protection systems
with respect to migrant workers, retired workers, their family members and students (regulation 1408/71 and its case law).
- Minimum harmonisation with regard to certain aspects of working conditions. Although Article 137 of the Treaty of the Eur opean Communities (TEC), which provides the legal basis for minimum harmonisation, also covers social security and the social pro tection of workers, there is no secondary leg islation related to the minimum harmonisa tion of social protection. On this aspect of Article 137, decision making is by unani mous voting.
- Open co-ordination with respect to social in clusion, the modernisation of pension sys tems and, as proposed by the Commission, health care systems.
Pros and cons of co-ordination
In this section we discuss the economic litera ture on the impact of economic integration on social protection levels. Do externalities pro vide grounds for social protection policy cen tralisation ?
Social protection in Europe
Effects of economic integration
Theoretically, economic integration can be beneficial as well as harmful to social protec tion systems. According to a well-known argu ment, economic development undermines tra ditional solidarities in family and local struc tures (Chassard and Quintin, 1993). In addi tion, increased labour mobility creates a need for employment-related insurance and broader solidarity networks. At the same time, higher income levels make it possible to develop a so cial security system with adequate protection levels. At any rate the funding of this type of system is facilitated. According to this line of reasoning, economic development reinforces the need for an extended system of social pro tection and makes it easier to fund it. And to the extent that European integration promotes economic development by reducing uncer tainty, lowering risk premiums and improving investment opportunities, it can contribute to the expansion of these systems. Intensified in ternational contact can also promote policy convergence.
One might argue though that economic inte gration may be harmful to national social se curity plans. Economic integration and an ef fective internal market can stimulate migra tion. Migration as a result of relative price sig nals is economically efficient. A different situation arises however if migration flows are provoked by differences in social protection le vels. In that case, an adverse selection problem arises; individuals who expect to be net benefi ciaries are attracted to countries with generous social programmes and net contributors are de terred by the high tax burden there. Conse quently, the social protection systems there are faced with increasing outlays and a narrowing financial base, ultimately resulting in lower protection levels (Sandmo 2001). This is a stan dard argument for centralising redistribution policies in an economic union, although it can be demonstrated that centralisation is not an inevitable consequence (Wildasin, 1991) and measures can be taken to limit and perhaps even eliminate the problem in practice.
Another problem may occur if lower levels of social protection translate into lower labour costs, in which case economic integration and higher transparency can damage the competi tive position of countries with more generous protection systems than elsewhere. They may
react by lowering their protection levels and thus set a race to the bottom into motion. So cial protection may indeed converge in the end, but only at a very low level of protection. This is usually called the so cia l dum ping or wage dum ping argument (Sinn 2003: 89). A more neutral term is p olicy com petition.
Both of these views on the effect of econom ic integration on social protection are echoed in Article 136 TEC, which states the objectives of European social policies. According to this article, the objectives of the improvement and convergence of living and working conditions and proper social protection will ensue from the functioning of the common market, fa vouring the harmonisation of social systems, and from minimum requirements for gradual implementation as stated in Article 137(2) TEC. Of course, the adaptation of EU m ini mum requirements is subject to the subsidiar ity principle. This means the Community only can take action if, by reason of scale or the ef fects of the proposed action, the objectives cannot be sufficiently achieved by the member states. 'The effects of the proposed action' per tain to the potential external effects, which in this case could arise due to either social secur ity migration or competition effects.
A closer look at the policy competition
argument
The possible effects of social security tourism on social protection systems are hardly dis puted in the literature. The main question is whether this poses major problems for the member states. We address this question in the last section. The social dumping or competi tion argument is more disputed in a theoreti cal as well as an empirical sense. We discern three themes in the literature:
- tax incidence. Does social spending increase the total wage costs or does it mainly affect the composition of the total wage costs? - competitiveness. Does the welfare state
pose a threat to competitiveness?
- spontaneous convergence. Are low stan dards, as argued by Sinn (2003), a necessary concomitant of a long adjustment process and will countries with lower social stan dards ultimately catch up with those with higher standards? Sinn argues that a prema ture harmonisation would delay the catch ing up process, resulting in large-scale mi
gration from the new members states to the old ones. He feels these migration flows would in turn put pressure on the social ex penditure in the old member states.
In the rest of this section we briefly discuss the first two points. The third one, sponta neous convergence, is more extensively ad dressed in the following section.
Tax incidence
Although employers usually pay part of the so cial security contributions, it does not go with out saying that their contributions automati cally make for higher nominal wage costs. The degree to which social spending results in higher wage costs generally depends on three factors (OECD 1990, Alogoskoufis et al. 1995, Alesina and Perotti 1997):
- the elasticities of the labour demand and supply curves. If the labour supply curve is highly inelastic (labour supply barely reacts to changes in wages) and the labour demand elasticity is within a normal range, then con tributions mainly affect the composition of the wage costs and not result in higher wage costs. In this case, employers can ultimately shift their contributions to their employees by offering lower wages;
- the degree to which employees view contri butions as insurance premiums instead of a tax on wages. This determines the incentive to change employment behaviour as a reac tion to higher contributions and the degree of tax shifting;
- the behaviour of trade unions. In the pre sence of trade unions, social contributions are paid in part by employers and thus in crease the labour costs even if the individual labour supply is perfectly inelastic. However, centralised comprehensive trade unions in ternalise the resulting adverse effects of higher wage costs on employment and thus moderate wage claims.
Because of differences in short- and long-term elasticities and trade union behaviour and pos sible differences in employee appreciation of social protection, it is hardly surprising that empirical studies do not have an unambiguous answer to the question of whether higher con tributions result in higher labour costs. On the basis of several studies, The OECD Jobs Study (OECD 1995: 247) concludes that increases in
taxation are often but not universally found to affect real wage costs. However, differential re sponses for different taxes should be inter preted with caution. Nickell (2003) concludes in his recent survey that there is probably some adverse long-term tax effect on labour costs. But even if this effect were substantial (which is not obvious), it does not necessarily mean increased policy competition puts down ward pressure on social spending. It d oes mean increased policy competition may force em ployees to pay a larger part of the social expen diture. From this perspective, policy competi tion mainly has distributional consequences.
Competitiveness
A number of studies have recently addressed the issue of whether welfare states can survive globalisation. They investigate the relations be tween economies' openness, competitiveness, and social spending levels (Rodrik 1998; Agell 1999; Krueger 2000; De Grauwe and Pollan 2003). Perhaps surprisingly, the studies typi cally observe a positiv e relation between the degree of openness and the level of social spending. On the average, countries more prone to external competitive pressures have higher expenditure ratios. There are a number of possible explanations for a positive relation between openness and social expenditure ra tios:
- open economies are usually small with cen tralised comprehensive unions that interna lise the possible adverse effects of social spending on labour costs and competitive
ness,-- open economies are generally rich. Richer countries can afford to spend more on social programmes. Moreover, social spending may be considered a luxury good with an income elasticity greater than one. This causes social spending to rise proportionally faster than income (see next section);
- since open economies are more prone to the ups and downs of the world market, the de mand for social protection is greater there; - adequate social protection may foster risk
taking, which can stimulate productivity and competitiveness.
This is not the appropriate place to discuss which explanation is best. Moreover, they are not mutually exclusive. The point we want to make is that the literature on tax incidence as
Social protection in Europe
well as on the relation between openness and social expenditure ratios raises serious doubts about the premises of the policy competition argument.
Social convergence?
The evidence surveyed above raises doubts about the validity of the policy competition ar gument. Further evidence can be accumulated by looking more directly at the developments over time of social spending and replacement rates. Is there spontaneous convergence? If so, is it towards a high or a low level ? What are the possible explanations?
Tests on convergence
Have social protection systems converged un der the influence of economic integration ?
Cornelisse and Goudswaard (2002) analyse whether social protection systems have actu ally converged or diverged in the past few dec ades. For this purpose, they use data on the gross replacement rates of unemployment ben efits and data on the share of the GDP spent on social benefits. The social security expendi ture ratio gives an indication of the financial effort to provide social protection, and replace ment rates are a measure of the benefits level and thus the generosity of the welfare pro grammes. A test on convergence is carried out using the standard deviation and coefficient of variation as yardsticks. They use the terms rela tive convergence or divergence when observing a drop or rise in the value of the coefficient of variation and the terms absolu te convergence or divergence when using the standard devia tion as criterion.2 Cornelisse and Goudswaard note a relative as well as an absolute conver gence of replacement rates for the EU coun tries in the past few decades. The coefficient of variation dropped by a third since 1980. Social expenditure ratios show a rather strong relative convergence (the coefficient of variation fell by some 30% since 1980), especially in the EU countries (the present member states), but also in the non-EU OECD countries. However, the EU countries do not exhibit absolute conver gence (the standard deviation hardly changed during the period of observation). In other words, the relative convergence they observe is the result of a rise in the average value of the
expenditure ratio.
Table 1, which is based on the most recent Eurostat data, also indicates a strong conver gence of relative as well as absolute social pro tection spending. From 1980 to 2001, the stan dard deviation of social spending declined by 60% and the coefficient of variation by 59%. Especially the Mediterranean countries, with rather low levels of protection in 1980, caught up rapidly in terms of social spending. This lar gely explains the rather strong social conver gence.
Dekker et al. (2004) estimate the /3-conver gence of social spending by regressing the growth of social spending on the initial level of expenditure. They find a /?-convergence of 4% a year in the 1980-1998 period. This means the difference between the national and the EU average declines by 4% a year.
But what about the indicators? They have several limitations. Statistics on social spend ing and benefits are known to be difficult to compare across countries. Countries use dif ferent definitions of social security and social protection. Perhaps the most important pro blems are related to differences in the public/ private mix in the provision of social
protec-Table 1 Social Protection* Expenditure as % of the GDP (EU 15) 1980 7990 2001 Austria 22.4 26.7 28.4 Belgium 28.0 26.4 27.5 Denmark 28.7 28.7 29.5 Finland 18.6 25.1 25.8 France 25.4 27.9 30.0 Germany 28.8 25.4 29.8 Greece 9.7 22.9 27.2 Ireland 2 0 .6 18.4 14.6 Italy 19.4 24.7 25.6 Luxembourg 26.5 22.1 21 .2 Netherlands 30.1 32.5 27.6 Portugal 12.9 15.2 23.9 Spain 18.2 19.9 20.1 Sweden 32.0 33.1 31.3 United Kingdom 21.5 23.0 27.2 Average 22.9 24.8 26.0 Standard deviation 3.75 2 .0 0 1.55 Coefficient of variation 0.25 0.13 0 .1 0
* Including sickness, health care, disability, old age, sur vivors, maternity, family, vocational guidance, unem ployment, housing and miscellaneous benefits. Source: Social Protection in Europe, Eurostat.
Table 2 Net Social Expenditure as % of the GDP, 1997 (ranked according to net social expenditure) Gross public social expenditure Gross total social expenditure
Net total social expenditure Sweden 31.8 34.8 27.3 Germany 26.4 28.6 26.1 Belgium 27.2 29.5 25.4 Denmark 30.7 32.0 23.5 Italy 26.4 27.8 22.7 Finland 28.7 30.0 22.1 Austria 25.4 27.0 2 2 .0 United Kingdom 21 .2 24.9 21 .8 Netherlands 24.2 29.1 21.5 Ireland 17.6 19.2 16.5 Average 26.0 28.2 22.8 Coefficient of variation 0.15 0.14 0.13 Standard deviation 4.02 3.98 2.8 8
Source: Adema (2001), Goudswaard and Caminada (2003).
tion and differences in the tax systems.
Adjusted indicators
The OECD has developed indicators to meas ure how much governments really devote to social spending, n et pu b lic so cia l expenditure, and the segment of an economy's domestic production that social benefit recipients draw on, n et total so cia l expen ditu re (Adema 2001). Benefits may be provided by either public or market institutions. In the latter case, the mar ket provisions may be regulated by the state in such a way as to make them equivalent to the public provisions. The various forms of social protection are not consistently included in the national statistics. A specific statistical pro blem is related to how the tax system treats so cial benefits. Benefits are taxable as a rule in some countries but not in others. Benefits can also be given as tax relief, for example a tax de duction for children. Table 2 presents figures on the net social expenditure as percentage of the GDP for 1997 for the EU countries data are available on.
The data clearly indicate that the adjust ments have an equalising effect on levels of so cial effort across countries. The standard de viation and coefficient of variation both exhi bit a decline.
Table 3 presents a comparison of the countries information is available for on the net replace ment rates of unemployment benefits.3 Only
the first period of unemployment is consid ered (social assistance included).
The net adjusted replacement rates appear to be much higher than the gross rates. Again, we
Table 3 Gross and Net Replacement Rates of Unem ployment Benefits in % , 1999 Index gross replacement rates Index net replacement rates Austria 41.7 69.5 Belgium 45.7 72.4 Czech Republic 2 2 .0 6 6 .8 Denmark 6 6 .0 80.8 Finland 54.0 81.0 France 59.0 75.3 Germany 37.0 67.5 Greece 41.3 47.1 Flungary 50.0 62.3 Ireland 35.0 51.4 Italy 60.0 45.5 Luxembourg 80.0 84.5 Netherlands 70.7 84.8 Poland 29.0 51.5 Portugal 65.0 83.0 Slovak Republic 40.0 77.8 Spain 63.0 74.8 Sweden 74.0 81.5 United Kingdom 17.3 53.6 Average EU 75 54.1 72.1 Coefficient of variation 0.32 0.17 Standard deviation 17.2 12.3
Source: OECD, Benefits and Wages, OECD Indicators, 2002, Goudswaard and Caminada (2003).
Social protection in Europe
calculate the standard deviation and coefficient of variation. The coefficient of variation dropped by 47 % and the standard deviation by 27 %.
To conclude, this analysis indicates that ac counting for private social benefits and the im pact of the tax system does indeed exert an equalising effect on the levels of social effort across member states. Unfortunately, there are no time series of the adjusted figures of social protection. However, the fact that differences in the adjusted protection levels have become so small suggests that social protection conver gence may have been even stronger than in earlier studies.
How to explain convergence?
In his recent book on the development of so cial spending and economic growth, Lindert summarises five leading forces that help ex plain the rise of social expenditure as well as cross-country differences in their rise (Lindert 2004: 188): democratisation, ageing popula tions, globalisation, the rise of income per ca pita, and international differences in the social affinities felt by middle-income voters. To ex plain the convergence of social expenditure ra tios in EU member states, the first four factors seem to be relevant. With respect to the first factor, there is the transition to democracy that Spain, Portugal and Greece underwent from the mid-1970s onwards. The governments not only needed to stabilise the countries politi cally, they also needed to respond to new and pressing social needs brought about by the co inciding political transitions to democracy and economic crises (Castles 1995; Guillen and Al varez 2000). Together with the ageing popula tions and rise of income per capita, this prob ably explains most of the rise in the social ex penditure ratios in these countries, and conse quently the convergence of social expenditure ratios in the EU 15 (see table 1 above).
Scharpf (1999: 177) observes a strong correla tion between the GDP per capita and the social expenditure ratios in the EU. He notes that the member states are strikingly alike in their re vealed preference for social spending. By and large, the richer member states do proportion ally more public social spending than the poorer ones. It seems that past patterns of overall social spending can largely be ex plained by changes in the ability to pay. As Scharpf observes, this is by no means a trivial
explanation, since it does not hold for the total set of industrialised OECD countries.
We have done a simple regression analysis with an update of Scharpf's data for 2000, and include the new member states. Figure 1 shows the results. Excluding the two outliers Ireland and Luxembourg from the sample, the strong correlation between the GDP per capita and social expenditure ratios noted by Scharpf is confirmed (R2 = 0.65).4 The extreme position of Ireland is mainly explained by its relatively young population. Ireland is the only country in the sample where pension payments do not constitute the largest part of social spending. The next section focuses on the position of the new member states.
New member states
In most of the new member states, social ex penditure ratios are lower than in most of the old ones. Does this give the new member states a competitive advantage? And will it set a race to the bottom in motion (e.g. Vaughan- Whitehead 2003)? Moreover, are the social pro tection systems in the old member states threa tened by migration from the new ones? Or will the social expenditure ratios in the new mem ber states catch up with the levels in the old ones?
Social expenditure ratios
The average social expenditure ratio in the new member states is seven percentage points below the average for the old ones. However, as figure 1 shows, there is considerable variation among the new member states. The Polish ex penditure ratio is almost nine percentage points higher than the Estonian one. In fact, the Polish, Hungarian and Slovenian expendi ture ratios are near the EU 15 average, and well above the Irish one. This is quite surprising, given the much lower GDP per capita in these countries. As figure 1 shows, there is a rather strong relation between the GDP per capita and the social expenditure ratio in the en larged EU. The fit of the trend line even im proves if the new member states are included in the sample. From this perspective, the social expenditure ratios in the new member states are in line with their income levels and in some countries like Poland even well above their income level (see Lindert 2004: 216-217). Table 3 also shows that the adjusted net
35 30 15 10 S w e d e n ■ France G e r m a n y ■ ■ D e n m a r k ..•••■ ■ ■ A u s tria G re e c e .. a m " t h e N e th e r la n d s S l o v e n i a " U K Bel8 iu n ’ . , / Ita ly PO and" H u n g a iy x - j p 0 rtu g a l S lo w a k ia ■ _ ... _ . . ______________2 Ü Ü______________________________________________— L u x e m b u rg ■ ■ La tv ia C zech R ep u b lic ■ L ith u a n ia ■ E s to n ia B Ire la n d 1 _____ 1 ___ L . . ___1 . I Ü Ü i - i v 0 25 50 75 100 125 150 175 200 gdp/capita (ppp, EU-15 = 100)
Figure 1 Social Expenditure Ratios and the CDP Per Capita (2000).
Sources: Social expenditure ratios EU 15: Eurostat, European Social Statistics, Social Protection, Expenditure and Re ceipts Data 1991-2003, Luxembourg 2003, p. 14 (cf. Table 1); New member states: Gesellschaft fiir Versicherungs- wissenschaft und -gestaltung, Study on the Social Protection Systems in the 13 Applicant Countries, November 2002 , p. 22; GDP per capita: EU15: European Commission, The EU Economy 2002, op. cit., pp. 286-287; New member states: Eurostat, Towards an Enlarged European Union, Key Indicators on Member States and Candidate Countries.
cement rates of unemployment benefits in the new member states do not differ very much from those in the EU 15 countries.
Pension reform in the new member states: a
case of social dumping?
The main reason why social spending in Po land, Hungary and Slovenia is relatively high has to do with their pension systems. For ex ample, pension expenditure in Poland is more than 14% of the GDP. Compared to the age structure of the population, these countries have relatively high numbers of retired people. The easy access to pension benefits emerged in three ways (see Barr 2001: 252). The general re tirement age under communism was low, the retirement age was very low for some privi leged groups like miners, and generous early retirement was frequently used as an exit route from the labour force in the transition period in the 1990s. As a result, the employment ra tios in general and for older workers in particu
lar are well below the figures for the old mem ber states (see table 4).
The new member states still have a rela tively young population compared to the old ones. However, mainly as a result of rapidly de creasing birth rates, the population as a whole is ageing in the new member states and age de pendency ratios will converge to the levels of the old ones. Combined with the high system dependency ratio as a result of the low effective retirement age, this could result in mushroom ing pension costs. If there are no reforms, pen sion expenditure in Poland for example would rise to more than a quarter of the national in come in 2050 (Lindeman et al. 2000). Against this background, the new member states led by Poland and Hungary are reforming their pension systems by de-indexing first-pillar pensions, raising the mandatory retirement age, and introducing second-pillar supplemen tary pension funds. Some authors (e.g. Vaugh- an-Whitehead 2003) consider these reforms a
Social protection in Europe Table 4 Employment Ratios and Unemployment Rates in New Member States in % , 2001
Employment ratio Unemployment Rate
Total Females 55-59 60-64 Cyprus 65.5 52.5 60.5 35.1 4.5 Czech Republic 64.9 56.8 50.2 16.9 n.b. Estonia 60.6 57.1 58.4 29.4 12.3 Hungary 55.9 49.4 33.7 7.6 5.7 Latvia 58.2 54.3 49.3 21.8 12.9 Lithuania 60.1 58.5 56.8 26.4 16.5 Poland 55.1 49.3 37.7 20.9 18.6 Slovakia 56.3 51.1 34.5 6.1 19.7 Slovenia 62.7 58.5 29.0 15.1 5.9 EU 15 64.0 54.9 51.9 22.6 7.4
Source: European Commission, The Social Situation in the European Union, Brussels 2003, p. 203. No data for
Malta.
form of social dumping. However, as is evident from the description of the pre-reform situa tion, the main aims of the pension reforms are to ensure the sustainability of the pension sys tems and stabilize the pension contributions.
Do low expenditure ratios imply lower tax
wedges?
Low employment ratios in the new member states mean a narrow basis for financing social spending. As a consequence, tax wedges (the part of gross labour costs that consists of social security contributions and income taxes) are comparable to the EU 15 average despite the lower average expenditure ratios (see table 5). This not only applies to high spenders like Po land, Hungary and Slovenia, but to low spen ders like the Baltic states as well (as to the low spenders see OECD 2003).
As table 5 shows, Hungary in particular has a very high tax wedge. Since a high tax wedge can discourage new entrants on the labour market and encourage the growth of the infor mal sector, a vicious circle can emerge with a high tax wedge causing decreasing official em ployment ratios and leading to still higher tax wedges and so forth (cf. Ederveen and Thissen 2004; European Commission 2003; United Nations 2004).
Is social expenditure too low in the new
member states?
We have given several reasons to be cautious about drawing conclusions from the fact that the average social expenditure ratios are lower in the new member states than the old ones:
- There are considerable differences among the new member states.
- The average social spending in the new member states is in line with their income levels and in certain countries even well above them.
- Mainly as a result of low employment rates related to a low retirement age, the basis for financing social expenditure is narrow, re sulting in relatively high tax wedges.
This is why we do not believe the lower social protection in the new member states will put pressure on the social protection in the old ones. This kind of scenario is unlikely anyway, since in economic terms, the new member states are very small compared to the old ones. We expect social spending in the new member states to converge to the levels in the old ones
Table 5 Tax wedge 1999
Poland 42.9 Hungary 52.6 Czech Republic 43.0 Slovak Republic 42.0 Slovenia 41.0 Estonia 40.0 Lithuania 39.7 Latvia 41.7 EU 15 weighted average 43.2 EU max (Belgium) 55.6 EU min (Ireland) 43.2
Adapted from Ederveen and Thissen (2004), p. 33. Source: Eurostat.
Tax wedge = employees' and employers' social secur ity contributions and personal income tax minus trans fer payments as percentage of gross labour costs.
as a result of demographic developments and rising income growth. The new member states mainly use lower company tax rates as instru ments in location competition. Although ag glomeration rents protect the old member states' higher company tax rates to a certain degree (Baldwin and Krugman 2002), it might put pressure on lowering company tax rates and thus on social protection spending as well (Tanzi 2000). However, this is indicative of a need for minimum company tax rates instead of minimum social protection levels.
It could also be argued that the new member states are reducing their social spending to meet with the entry criteria for the third phase of EMU. The budgetary criteria for admission to the euro-zone have been criticised in the past as being overly stringent, forcing govern ments to cut down on their social spending. However, judging from the experience of the current euro-zone members, there is no evi dence that budgetary consolidation in the 1990s, motivated by a desire to meet with the Maastricht criteria on debts and deficits, has resulted in substantial cuts in social spending. As table 1 shows, in fact social expenditure ra tios increased in the 1990s, even in countries like Italy, which started with a very high gov ernment deficit. The reason why is that most countries mainly used temporary tax increases and privatisation proceeds to consolidate their budgets (see Fatas et. al. 2003: 35-37). More over, once the financial markets started to be lieve the member states would join the euro, the risk premiums on interest rates rapidly de creased, leading to huge savings on interest ex penditure on the public debt. In connection with this, we do not believe sound public fi nance policies would necessarily hurt social policy spending. On the contrary, by cutting back interest rate expenditure, a low debt pol icy might contribute to the sustainability of social protection systems in the long run.
Social security tourism from new member
states?
The other channel through which EU enlarge ment could put pressure on social security sys tems in the old member states is migration. More specifically, higher benefits in the old member states could stimulate social security tourism, which could force the old member states to lower their benefits. However, we do
not think this scenario is very likely. Despite obvious benefit level differences between the southern and northern member states, we do not observe social security tourism in the old member states. One reason for this may be that EU citizens do not have unconditional rights to social assistance, unemployment benefits or mandatory health insurance in other member states. Residence rights for non-workers are usually conditioned on the ability to be self- supporting and not be a burden on the social security system in the host member state. Sinn (2003) argues that the Constitution drafted by the European Convention will give EU citizens an unconditional right to social welfare in other member states. He argues that this will encourage mass east-to-west migration flows and force the old member states to lower their social protection. We feel this argument is based on a misconception about the Constitu tion, which explicitly denies citizens new rights (Van Riel 2003, Geelhoed 2003). Court of Justice case law on the concept of European citizenship also does not prohibit member states from putting certain limitations on the access to their social protection systems to pre vent social security tourism.5 According to some authors (e.g. Bertola et al. 2001), these limitations hinder the emergence of a truly European labour market since they restrict the mobility of the unemployed. This is why the harmonisation of social minimum standards is felt to be called for. However, this efficiency ar gument in favour of social harmonisation fails to convince us for two reasons. Firstly, it has not been established that restrictions on the re sidence rights of the unemployed are an impor tant reason for their low mobility. Other im portant factors are cultural differences and m o bility costs. In addition, the unemployed are not very mobile within their own member states either. Secondly, the argument fails the proportionality test. Do we really need to har monise the social minimum standards to make job searching in other member states possible?
Conclusions
This article addresses the question of whether we need to accelerate social convergence in the enlarged EU by setting common minimum
Social protection in Europe
standards for benefit levels. Is there a need for the harmonisation of social protection sys tems, perhaps because of the increased danger of a social race to the bottom ? We have theore tically and empirically analysed the social dumping hypothesis and conclude that the case for harmonisation is not strong. There are no clear signs of a social race to the bottom. On the contrary, several indicators show that in the past few decades, protection levels have converged to higher averages. This process of social convergence was induced by the eco nomic integration process.
The new member states have lower average protection levels than the EU 15, although the differences are not that large, but we do not ex pect this to put pressure on the protection le vels in the old member states. It is important for the new member states to keep their labour costs low so as to be able to compete on the in ternal market. Once economic catching up has taken place, they will be able to develop a more mature social protection system. Premature social harmonisation would be detrimental to the economic convergence process. Welfare states could also be threatened by increased migration flows. However, we feel this threat is often overestimated. Citizens do not have un conditional rights to social benefits in other countries and other factors such as cultural dif ferences and mobility costs also restrict mass social tourism. In any case, harmonisation would not be the answer to this problem either.
Although we conclude that the EU does not need social benefit harmonisation, we do ac knowledge the importance of the social di mension of the EU in a broader sense for eco nomic as well as political reasons (legitimacy of the integration process]. The member states are faced with the common challenge of in creasing the sustainability of their social pro tection systems, which are endangered by var ious developments, especially the ageing popu lations. The method of open co-ordination can help implement the reforms. From an econom ic point of view, some economies of scale can be grounds for this co-ordination. Best prac tices and policy learning can be disseminated, and peer pressure and peer review can strengthen the commitment of member states to common or national objectives. However, open co-ordination has not yet been very ef
fective. The essential problem is that the feed back to national policy-making and conse quently the implementation is insufficient. We suggest a much stronger integration of the ac tion programmes based on common EU objec tives and guidelines in the national policy pro cess. This also means national parliaments should devote more attention to these pro grammes and evaluate their implementation more systematically. More effective EU social policy co-ordination can contribute to the sus tainability of the social protection systems of the member states and modernisation of the European social model, which is an essential part of the Lisbon process.
Notes
1 Leibfried and Pierson (1996:196) summarize the key implications of Regulation 1408/71 and its case law as follows: 1) A member state may no longer limit social benefits to its citizens. 2) A member state may no longer insist that its bene fits only apply to its territory and are thus only consumed there. 3) A member state is no longer entirely free to prevent other social policy re gimes from directly competing with the regime it has built on its own territory (e.g. in the case of posted workers). 4) Member states no longer have an exclusive right to administer claims to welfare benefits.
2 One property of the standard deviation is that its value rises with the average value of the data set it is applied to. The coefficient of variation is defined as the standard deviation divided by the average value.
3 The calculation of net replacement rates differs in several ways from the calculation of gross re placement rates (see OECD 2002). Taxes and so cial security contributions on earnings and ben efits are taken into account. Moreover, net repla cement rates capture the effect of family-related benefits for children. Housing benefits are also included in net replacement rates.
4 Without the new member states, the trend line is: (y is expenditure ratio, x = GDP per capita): y = 0.11‘x +16 (R2 = 0.30). Including the new member states improves the fit: y = 0.13 *x+14 (R2 = 0.65).
5 See for example the conclusion of Advocate Gen eral Geelhoed in theTrojani case (C-546/02).
References
Adema, W. (2001), Net Social Expenditure: 2nd ed i
tion, Labour Market and Social Policy - Occasional
Papers No. 52, Paris: OECD, 2001.
Agell, J. (1999), On the Benefits from Rigid Labour Markets: Norms, Market Failures and Social In surance, The Econom ic Journal 109: F143-F164. Alesina A. and R. Perotti (1997), The Welfare State
and Competitiveness, American Econom ic Re
view, vol. 87: 921-939.
Alogoskoufis, G., C. Bean, G. Bertola, D. Cohen, f. Dolado and G. Saint-Paul (1995), Unemployment:
Choices for Europe, CEPR, Monitoring European
Integration 5, London.
Atkinson, A.B. (2002), Social Inclusion and the Eur opean Union, in Journal o f Common Market Stu
dies, 40,4: 625-643.
Baldwin, R. and P. Krugman (2002), Agglomeration,
Integration and Tax Harmonisation, CEPR Dis
cussion Paper No. 2360, London.
Barr, N. (2001), The Welfare State as a Piggy Bank: In
formation, Risk, Uncertainty, and the Role o f the State, Oxford, Oxford University Press.
Bertola, G., T. Boeri and G. Nicoletti, (eds), Welfare
and Employment in a United Europe, MIT Press,
Boston.
Castles, F.G. (1995), Welfare State Development in Southern Europe, Western European Politics, 18: 291-313.
Cornelisse, P.A. and K.P. Goudswaard (2002), On the Convergence of Social Protection Systems in the European Union, International Social Secur
ity Review, 55, 3: 3-17.
Chassard, Y. and O. Quintin (1993), 'Towards a Con vergence of Social Policies', in J. Berghman and B. Cantillon (eds), The European Face o f Social Se
curity: Essays in Honour o f H. D eleeck, Avebury,
Aldershot, 337-355.
De Grauwe, P. and M. Polan (2003), G lobalisation
and Social Spending, CESifo Working Paper No.
885.
Dekker, P. et al. (2004), Sociaal Europa - Europese
Verkenning, Bijlage bij Staat van de Europese Unie
2004,The Hague: CPB/SCP.
Ederveen, S. and L. Thissen (2004), Can Labour Mar
ket Institutions Explain Unemployment Rates in New M ember States1 CPB Document No. 59, The
Hague: CPB.
European Commission (2003), Progress on the Im
plem entation o f the Joint Assessment Papers on Employment Policies in Candidate Countries,
COM (2003), 37, Brussels, 30 January 2003. Fatas, A., J. von Hagen, A. Hughes Hallett, R.R.
Strauch and A. Sibert (2003), Stability and
Growth in Europe: Towards a Better Pact, Lon
don: CEPR.
Geelhoed, L.A. (2003), Een Europawijde Unie: een grondwet zonder staat, SEW 9: 284-310.
Goudswaard, K.P. and C.L.J. Caminada (2003), Con vergence of Social Protection Reviewed, in A. P. Ros and H.R.J. Vollebergh (eds), O ntw ikkeling en
overheid, The Hague: Sdu, 97-106.
Hutsebaut, M. (2003), Social Protection in Europe:
A European Trade Union Perspective, Interna
tional Social Security Review, 56,1: 53-74.
Krueger, A.B. (2000), From Bismarck to Maastricht: The March to European Union and the Labor Compact, Labour Econom ics 7 :117-134.
Leibfried, S. and P. Pierson (1996), 'Social Policy', in: H. Wallace and W. Wallace (eds), Policy- Making in
the European Union, Oxford: Oxford University
Press, 185-207.
Lindeman, D , M. Rutkowski, O. Sluchynkyy (2000), The Evolution o f Pension Systems in East
ern Europe and Central Asia: Opportunities, Con straints, D ilem m as and Emerging Practices, Wa
shington DC:TheWorld Bank.
Lindert, P.H. (2004), Growing Public: Social Spend
ing and Econom ic Growth Since the Eighteenth Century, Cambridge: Cambridge University Press.
Mooij, R. de and P. Tang (2003), Four Futures o f Eur
ope, The Hague: CPB.
Nickell, S. (2003), Employment and Taxes, CESifo Working paper no. 1109, Munich: CESifo.
OECD (1995), The OECD Jobs Study, Evidence and
Explanations, Paris: OECD.
OECD (2002), Benefits an d Wages. OECD Indicators, Edition 2002, Paris: OECD.
OECD (2003), Labour Markets and Social Policies
in the Baltic Countries, Paris: OECD.
Rodrik, D. (1998), Why do more open economies have bigger governments? Journal o f Political
Economy, 166, 5: 997-1032.
Sandmo, A. (2001), Globalisation and the Welfare State: More Inequality, Less Redistribution? in D. Pieters (ed.), European Social Security and Global
Politics, The Hague: Kluwer Law International,
45-60.
Scharpf, F. (1999), Governing in Europe: Effective
and Dem ocraticI Oxford: Oxford University
Press.
Sinn, H.-W. (2003), The New Systems Competition, Blackwell Publishing.
Sinn, H.-W. and W. Ochel (2003), 'Social Union, Con vergence and Migration', Journal o f Common
Market Studies, 41, 5: 869-896.
Tanzi,V (2000) G lobalization and the Future o f So
cial Spending, IMF Working Paper, WP/00/12,
Washington DC: International Monetary Fund. United Nations (2004), Economic Survey o f Europe
2004, No. 1, New York and Geneva: UN, Econom
ic Commission for Europe.
Van Riel, B. (2003), Social Security Tourism a Red Herring, letter to the editor, Financial Times, 19 February, 2003.
Vaughan-Whitehead, D.H. (2003), EU Enlargement
versus Social Europe? The Uncertain Future o f the European Social Model, Cheltenham: Edgar El
gar.
Wildasin, D. (1991), Income Redistribution in a Common Labour Market, American Economic
