Expenditure on social protection in the European Union (EU) member states has been increasing rapidly over the last decade. To cover the increasing expenses, the countries need to find ways to increase revenues. Social protection financing systems and structure of financing vary across countries, but all of them use mainly two sources for financing: general government contributions and social tax revenue. The aim of this paper is to study the development of the structure of social protection financing at the main contributor level over the last decade, defining the trends that characterize the changes. We concentrate on convergence analysis of the structure of social protection financing, which is an important but, so far, insufficiently studied issue. (JEL H55, H53, P52)
EU member states make constant efforts to increase revenues to finance social protection. Social protection expenditures are financed from various sources, the most important of which, in most of the countries, is social tax revenue. It is extremely important to prevent the rise of tax rates in all countries where the main source of financing is social tax revenue. To cover the growing social protection expenditure, the structure of financing has been changed in the last decade, in which processing the share of government contributions have increased. In 1993, on average, 63.0% of the social protection expenditures in EU were covered from social tax revenue (contributions paid by employers and insured persons) and 32.8% from government contributions (public financing). In 2001, this ratio was 60.5% and 36.0%. The tendencies have been in different directions across countries.
This paper seeks to study the development of the structure of social protection financing at the main contributor level over the last decade, defining the trends that characterize the changes. We concentrate on convergence analysis of the structure of social protection financing, which is the most important part of social protection systems but a relatively stable area of social policy in EU countries. We study whether convergence or divergence has occurred in social protection financing structures in EU countries in the period 1993-2001, while the social protection expenditures have converged [Puss et al., 2003]. We analyze the presence of [sigma]-convergence, on the basis of linear and non-linear regression, the presence of absolute [beta]-convergence, and the speed of convergence in the structure of social protection financing in the EU.
In the EU member states, social protection has been an area of national competency. European Community legislation has therefore, purposefully avoided harmonization of national social security legislation of Member States. The Charter of Fundamental Social rights of 1989 made reference to the principle of subsidiarity and thereby, recognized the Community as an actor in social security. Subsidiaries conditionally give the Community the competency to act if objectives set out have not been fulfilled in a satisfactory manner at the national level [De la Porte, 2000].
The Social Charter also opened the debate on the setting of common objectives of social policies, where policies are understood as the manner in which the social protection systems develop. The Commission was the initiator of this process and believed that by setting common goals, a convergence of policies of social protection would be reached [De la Porte, 2000]. At the end of 1990s, due to the multiple factors including the European employment strategy, aging of population, transformation of the macroeconomic environment in the adaptation to the EMU, and enlargement of the EU, social protection has become more important in the EU. Despite all these developments, the organization of the social protection systems remains an area of national competency. At the national level, the efforts to balance social protection resources and expenditure can be made.
The issue of financing is the subject of political discussions, both in the EU as a whole and in individual countries. Social protection expenditures are financed from four sources. For financing social protection measures, most countries use means collected with the help of a special tax, i.e. social tax. This tax rate and principles of taxation vary considerably across countries. In all European countries, this is the payroll tax paid by both employees and employers. Additionally, some social protection expenditures are also covered by the public sector from general government tax receipts, and to a small extent, from various other finances.
First, we discuss changes in the structure of two main sources of financing. During the study period, the share of social tax contributions in the structure of financing has diminished and the state's share have increased in the EU on average (see Figure 1). Compared with 1993, the share of social tax paid by employers has decreased by 0.7 percentage points in the study period. The share of social tax contributions paid by insured persons has also diminished by 1.8 percentage points. This has been compensated primarily by an increase in government contributions, which have increased 3.2 percentage points. There are other finances that come from different sources across countries, but they are not discussed in this paper.
Differences in the institutional and organizational structure of social protection systems between EU countries are revealed in the structure of social protection financing. Though the development of the financing structure seems quite stable at the weighted average level, the country specific changes are significant, and in order to examine the convergence behavior, we have to concentrate on the country specific observations and a classical approach to the convergence analysis.
On the basis of the structure of social protection financing in EU member states (data of the last two years), we can divide EU member states into four groups. The first group comprises the countries where more than 65% of the expenditure is covered from social tax receipts. This group includes Belgium, Germany, Spain, the Netherlands, and France. The second group is formed of countries where the government contributions are the principal part (about 60%). This group includes Denmark and Ireland.
[FIGURE 1 OMITTED]
The Government's contribution is higher than the EU average. This is also true for the U.K., Sweden, Luxembourg, and Finland. These countries form the third group. In the rest of the countries, social tax receipts account only for a slightly higher share than government contributions.
The biggest changes, compared with 1993, occurred in the financing structures of countries in the first and second group, i.e. in the countries that most differed from the EU average level. Additionally we analyze changes in the structure of social protection financing in EU member states compared to EU average and ascertain the linear trends that characterize them. The analysis indicates that the share of social tax receipts in social protection financing has diminished in six countries, with the highest share of social tax being in France, Germany, Italy, Austria, Luxembourg, and Spain. Only these countries have introduced measures to limit social protection expenditure and increase general tax contributions in social protection financing. In Denmark, the share of social tax has increased dramatically in financing in recent years. The establishment of a new employees' health insurance tax causes this. In Belgium, Greece, and Netherlands, the share of social tax has also increased, although, it was initially already high in these countries.
Methodology and Data
According to the economic theory of convergence, the economic development level of less developed countries should approach the level of more advanced countries with the same economic resources or fundamentals. Socio-economic convergence is mainly discussed in the context and on the basis of two main economic growth theories: neoclassical and endogenous. Two main concepts of convergence are used in classical literature of growth theory: [sigma]-convergence and [beta]-convergence [Quah, 1996; Sala-i-Martin, 1996]. According to Sala-i-Martin, the concept of [sigma]-convergence can be defined as follows. A group of economies are converging in the sense of [sigma] if the dispersion of their real per capita GDP levels tends to decrease over time [Sala-i-Martin, 1996, p. 1020]. A reduction of the standard deviation or coefficient of variance of indicators indicates a reduction of the difference or the presence of [sigma]-convergence.
The test for the presence of [beta]-convergence that [beta]-convergence exists if poor economies tend to grow faster than rich ones posits [DeLong, 1988; Barro and Sala-i-Martin, 1991, 1992a, 1992b; Boyle and McCarthy, 1997]. In the real world, the economies may differ not only in their level of technology but also in other things, such as initial level of human capital or various government policies.
The literature makes a distinction between absolute (unconditional) and conditional [beta]-convergence. Absolute [beta]-convergence pertains to the coefficient [beta] of the bivariate equation. This is based on the assumption that all countries in a sample converge to the same steady state. Conditional [beta]-convergence pertains to the coefficient [beta] of the socioeconomic level variable in an equation that includes additional explanatory variables reflecting differences across countries.
We used the following equation to estimate absolute [beta]-convergence:
[1/T]ln([Y.sub.iT]/[Y.sub.i0]) = [alpha] + [beta]ln[Y.sub.i0] + [[epsilon].sub.iT], (1)
where the left-hand side is the average annual growth rate of the health care expenditure in country i at time T, [alpha] is a constant term, and [epsilon] is the error term. The condition for [beta]-convergence is the test that [beta]<0.
For the estimation of the speed of convergence we used the following equation:
[1/T]ln([Y.sub.iT]/[Y.sub.i0]) = [alpha] - [(1 - [e.sup.-bT])/T]ln[Y.sub.i0] + [[epsilon].sub.iT], (2)
where b is the rate of convergence.
We know that using this method may cause upward biases of [beta], but an important decision in making the choice was a relatively short time period. For testing the validity of the results, we used pooled OLS, Re-Weighted Least Squares, and Least Trimmed Squares methods.
We assess changes in the structure of social protection financing in the EU member states by the share of different contributors in social protection financing. We analyze the presence of [sigma]-convergence and [beta]-convergence in four major contribution groups for covering social protection expenditure: employers' social contributions, social contributions paid by the protected persons, general government contributions, and other receipts. We used harmonized, cross-sectional data of social protection financing presented by the Statistical Office of the European Communities (Eurostat). Our sample covers the period of 1993-2001 and the EU-15 member states.
The presence of [sigma]-convergence has been assessed for the level of structural elements in social protection finance, discussing first the changes in both social tax receipts and government contributions.
In the case of both factors, we can detect, based on the above analysis, the presence of [sigma]-convergence, indicating convergence of the financing structures of social protection expenditure. The analysis in Table 1 shows that the variation coefficient (indicator of [sigma]-convergence) of social tax receipts decreased over the years 1993-2001 (from 0.301 to 0.210). Table 2 contains similar data for the government contributions. The variation coefficient of government contributions also decreased in this period (from 0.438 to 0.336). Earlier studies on social protection financing demonstrated that in the period from 1980 to 1995, a mild [sigma]-convergence was found in the financing structure [Hagfors, 1999, 2000]. We found that the convergence in two structural elements of financing--social tax and government contributions--was noticeable in the period 1993-2001 (based on [sigma]-convergence). The ratio of maximum to EU average diminished for both indicators: social tax from 1.213 to 1.102 and government contributions from 2.485 to 1.785. The growth of minimum indicators increased in the case of social tax from 0.184 to 0.502 but diminished in the case of government contributions from 0.607 to 0.453.
Social tax is paid by both employers and insured persons (employees). The tendencies that have occurred in these proportions are described in Tables 3 and 4.
In terms of these indicators, we can state that the changes that have occurred in these countries have conduced to convergence between them. The analysis indicates that the variation coefficient of employers' contributions decreased in the years 1993-2001 from 0.179 to 0.114. The variation coefficient of contributions by insured persons decreased in this period from 0.502 to 0.334. The ratio of maximum to EU average diminished for both indicators: employers' contribution from 1.284 to 1.093 and contribution by insured persons from 1.791 to 1.627. The ratio of minimum indicators to EU average increased from 0.175 to 0.240 and from 0.085 to 0.420.
Following [Sala-i-Martin, 1996], we regress linearly the annual growth rate of the financing share by contributors with the initial level of the share at the beginning of the period. If the slope coefficient ([[beta].sub.1]) is negative, we agree that absolute convergence in social protection financing share exists (Table 5). In the same way, we can estimate the speed of convergence using a non-linear model. The positive value of [beta] indicates convergence, and the speed of convergence grows with the increasing value of [beta].
According to our findings presented in Table 5, there has been absolute convergence in the shares of social protection financing, and each of these has been converging towards its average. The presence of absolute convergence ([[beta].sub.1] < 0) for all shares of social protection financing indicates dependence of growth in the respective share on the initial level of the indicator. Our estimation gave statistically significant coefficients for all social protection financing contributor groups over the period 1993-2001. The relation is stronger for protected persons (Adj. [R.sup.2] = 0.81) (Table 5). The [[beta].sub.1] coefficient is negative and statistically significant with value [[beta].sub.1] = -0.082 and 95% confidence interval -0.106 < [[beta].sub.1] < -0.059 and also [[beta].sub.1] < 0 for the share of protected persons in social protection financing (for the share of employers [[beta].sub.1] = -0.025 and 95% confidence interval -0.044 < [[beta].sub.1] < -0.005). On the whole, the results satisfy the strong conditions for convergence in social protection financing shares for protected persons and moderate for employers' and general government share (but the share of general government has lower statistical significance) across 15 EU member states over the time period 1993-2001. Our findings are the same with those presented by Hagfors [1999, 2000], where he concluded that an absolute convergence of the structure of social protection financing occurred in the EU-14 countries during the period 1980-1995.
Figure 2 helps visualize the results. The three panels represent the share of employers' contribution, contribution of the protected persons, and general government (the initial share level is on the horizontal axis and the annual rate of growth on the vertical axis).
According to the definition of absolute convergence, for convergence to exist, the regression line must be downward sloping. Figure 2 demonstrates also that there was a strong absolute convergence in the protected persons financing share in 1993-2001, as the line shows a steep downward slope. The slope is mostly due to changes in the social protection financing structure in Denmark and Sweden. Without these two countries where the share of protected persons was very low compared to EU average (20% and 9%, respectively) and growth fast in that period, the coefficient [[beta].sub.1] would have fallen to the level of -0.016 but still would have been statistically reliable. We should mention that Denmark has nearly caught up with the EU average, while in Sweden, the share of protected persons financing is way below that level. The share of employers' and government's contribution shows moderate convergence, as the line is slightly downward sloping. The share of other sources of financing indicates that there has been moderate convergence in 1993-2001.
The speed of convergence has been rather low with the exception of the financing share of protected persons, which has experienced a quite rapid convergence. Starting from 1993, the share of protected persons would take approximately 11 years for one-half of the difference in the share of protected persons between member states and EU average to disappear at a 6.3% annual rate of convergence (the parameter [beta] estimating the speed of convergence indicates that social protection expenditures were converging in the period 1993-2001 by 6.3% annually). The speed of convergence was particularly low for the financing share of employers, and it would have taken 31 years for one-half of the difference in the share of employers between member states to catch up to the EU average, due to the annual rate of 2.2%.
[FIGURE 2 OMITTED]
This paper examined the evidence and degree of cross-country convergence in the financing structure of social protection expenditure in the EU during the period 1993-2001. We discussed changes in the proportions between financing from social tax receipts and government contributions and detected some harmonization in the structure of social protection financing in the study period. The analysis indicated that in the countries where social tax receipts were predominating at the beginning of the period, their shares have diminished and government contributions have increased. Our findings support, in general, the presence of [sigma]-convergence and absolute [beta]-convergence for social protection financing structure. The convergence has been strong for the convergence in social protection financing shares by protected persons and moderate for employers and general government across 15 EU member states over the time period 1993-2001.
Barro, Robert J.; Sala-i-Martin, Xavier. "Convergence across States and Regions," Brookings Papers on Economic Activity, 1, 1991, pp. 107-82.
Barro, Robert J.; Sala-i-Martin, Xavier. "Convergence," Journal of Political Economy, 100, 2, April 1992a, pp. 223-49.
Barro, Robert J.; Sala-i-Martin, Xavier. "Regional Growth and Migration: A Japan-United States Comparison," Journal of the Japanese and International Economies, 6, December 1992b, pp. 312-46.
Boyle, Gerry E.; McCarthy, Thomas G. "A Simple Measure of [beta]-Convergence," Oxford Bulletin of Economics and Statistics, 59, 1997, pp. 257-64.
De la Porte, Caroline. "Is There an Emerging European Consensus on Social Protection?," European Trade Union Institute Yearbook, 2000, pp. 309-40.
DeLong, Bradford J. "Productivity Growth, Convergence and Welfare: Comment," American Economic Review, 78, 5, December 1988, pp. 1138-54.
Hagfors, Robert. "The Convergence of Financing Structure 1980-1995," Financing Social Protection in Europe, Helsinki: Ministry of Social Affairs and Health, 1999, pp. 37-66, Publications 21.
Hagfors, R. "EMU Convergence and the Structure of Social Protection Financing," The Year 2000 International Research Conference on Social Security, Helsinki, 2000, p. 36.
Puss, Tiia; Viies, Mare; Maldre, Reet. "Convergence Analysis in Social Protection Expenditure in the European Union," in Essays in Estonian Transformation Economics, Ulo Ennuste and Lisa Wilder (eds.) Tallinn, Estonia: Uhiselu, 2003, pp. 123-46.
Quah, Danny T. "Twin Peaks: Growth and Convergence in Models of Distribution Dynamics," The Economic Journal, 106, 437, 1996, pp. 1045-55.
Sala-i-Martin, Xavier. "The Classical Approach to Convergence Analysis," Economic Journal, 106, 1996, pp. 1019-36.
TIIA PUSS*, MARE VIIES*, AND KAIE KEREM*
*Tallinn University of Technology--Estonia. This paper was presented at the 57th IAES Conference in Lisbon, March 10-14, 2004. The authors would like to thank the discussants in that session for their helpful suggestions. This paper was prepared with the support of the Estonian Science Foundation research grant. We wish to thank Reet Maldre for the assistance.
TABLE 1 The Evidence of [sigma]-Convergence in Social Tax Contributions Year Arithmetical Mean Standard Deviation Coefficient of Variation 1993 0.870 0.262 0.301 1994 0.878 0.234 0.266 1995 0.876 0.227 0.259 1996 0.881 0.223 0.253 1997 0.894 0.214 0.239 1998 0.915 0.204 0.223 1999 0.922 0.200 0.217 2000 0.929 0.196 0.211 2001 0.935 0.196 0.210 TABLE 2 The Evidence of [sigma]-Convergence in Government Contributions Year Arithmetical Mean Standard Deviation Coefficient of Variation 1993 1.182 0.518 0.438 1994 1.165 0.490 0.421 1995 1.191 0.472 0.396 1996 1.175 0.455 0.388 1997 1.159 0.434 0.375 1998 1.103 0.383 0.347 1999 1.090 0.373 0.343 2000 1.077 0.364 0.337 2001 1.061 0.356 0.336 TABLE 3 The Evidence of [sigma]-Convergence in Employers' Contribution Year Arithmetical Mean Standard Deviation Coefficient of Variation 1993 0.885 0.159 0.179 1994 0.887 0.158 0.178 1995 0.892 0.172 0.192 1996 0.899 0.169 0.188 1997 0.909 0.163 0.180 1998 0.927 0.150 0.162 1999 0.927 0.133 0.143 2000 0.931 0.101 0.108 2001 0.937 0.107 0.114 TABLE 4 The Evidence of [sigma]-Convergence in Contribution by Insured Persons Year Arithmetical Mean Standard Deviation Coefficient of Variation 1993 0.845 0.424 0.502 1994 0.863 0.400 0.464 1995 0.850 0.352 0.414 1996 0.852 0.343 0.402 1997 0.872 0.340 0.390 1998 0.896 0.303 0.338 1999 0.915 0.313 0.343 2000 0.925 0.323 0.349 2001 0.942 0.315 0.334 TABLE 5 Absolute [beta]-Convergence in Social Protection Financing Structure and Speed of Convergence in EU in 1993-2001 Adj. Contributors [[beta].sub.0] [[beta].sub.1] [beta] [R.sup.2] Employers Coefficient 0.093** -0.025** 0.022** 0.36 Std. error 0.032 0.009 0.008 Lower 95% 0.025 -0.044 0.008 Upper 95% 0.161 -0.005 0.037 Protected persons Coefficient 0.250*** -0.082*** 0.063*** 0.81 Std. error 0.031 0.011 0.007 Lower 95% 0.182 -0.106 0.050 Upper 95% 0.319 -0.059 0.076 General government Coefficient 0.105* -0.029* 0.026* 0.23 Std. error 0.054 0.015 0.012 Lower 95% -0.011 -0.062 0.002 Upper 95% 0.222 0.003 0.050 *p < 0.1 **p < 0.05 ***p < 0.01…