MPIfG Working Paper 97/7, July 1997


Employment and the Welfare State: A Continental Dilemma


by Fritz W. Scharpf mail


Fritz W. Scharpf Max Planck Institute for the Study of Societies, Cologne




Estimates of the comparative health of the North American and Western European economies and societies have had their fashion cycles - from Servain-Schreiber's warnings that Europe was falling behind, rather than catching up with, American technological leadership in the 1960s, to European exasperation over American trade and budget deficits in the 1970s, to anxieties over Eurosclerosis in the early 1980s and over the American loss of international competitiveness in the late 1980s. Presently, by all accounts, the sick man is again Europe, with higher unemployment and much lower rates of job creation over the last two decades or so. The main problem is a rising level of long-term unemployment that mainly affects unskilled workers and, in most countries, young job seekers with low levels of schooling.


For an explanation, most commentators see little need to search beyond the usual suspects - institutional rigidities, union power, and the burdens of the welfare state. In an age of intensified global competition, so it is argued, government regulations and collective-bargaining agreements can no longer be considered "beneficial constraints" but have become fetters that prevent European firms from achieving the flexibility and innovativeness that allow American firms to compete successfully at the cutting edge of high-tech markets. In addition, the European economies are laboring under the dead weight of bloated public sectors that are claiming more than 50 percent of GDP in Sweden and Denmark, more than 40 percent in France, Germany and other Continental countries, as compared to less than 30 percent in the United States and Japan - and of overextended welfare states with overly generous rates of income replacement that have raised reservation wages and reduced the incentives to work.



The Measure of Success: Employment Ratios


However, as you surely will have guessed, I am not convinced that things are quite as simple as the conventional wisdom would have them. If the employment performance of different countries is to be evaluated and explained, there is, first, a need for valid criteria. For this, the usual reference to unemployment figures will not do. They do not include persons on disability pensions, in early retirement and other forms of paid non-work, and they are notoriously subject to political manipulation - the British Conservatives are said to have changed unemployment definitions more than thirty times during their period in office, each time reducing the number of the registered unemployed. More importantly, the rate of unemployment is defined by reference to the size of the "active population" which is, of course, strongly affected by factors on the supply side of the labor market. Thus, the willingness of married women to enter the labor market may be as much affected by the separate or joint taxation of spouses' incomes or the availability of day care as it is by the availability of jobs.


Compared to unemployment rates, employment figures and their changes over time seem to be a much better indicator of comparative performance - but they also are affected by changes on the supply side: larger populations imply more jobs. Even when employment figures are normalized by reference to the population of working age (15 to 64), comparability suffers from differences in working time and in the share of part-time employment (which happens to be unusually high in the Netherlands). With this caveat in mind, however, employment/population ratios still seem to be the most valid indicators of relative employment performance for which internationally comparative data are readily available in OECD publications (OECD 1996a). I will use these throughout.


Similarly, the causal relationship hypothesized by the conventional wisdom is not fully captured by the usual reference to the share of taxes and social security contributions in GDP, since a large (but variable) part of public expenditures is actually meant to increase the productivity and hence the international competitiveness of national economies. Instead, I will refer to the share of total social expenditures in GDP (OECD 1996b) which, better than any other internationally comparable indicator, should indeed reflect the "dead weight" of the welfare state on the economy that is presumed to explain the poor European employment performance.

Taking the latest available OECD data on employment ratios for 1995, and on total social expenditures for 1993, it is certainly not obvious that the conventional wisdom has much explanatory power (OECD 1996a). It is true that the United States, at 73 percent, and some other countries with very low shares of social spending had very high employment ratios, and it is also true that the German score, at 65.1 percent, was significantly lower, closely followed by the Netherlands (with all those part-time jobs) at 64.3 percent. France, at 59.5 percent, was even further behind, and in Belgium only 55.7 percent of the working-age population were actually working. Ireland and the Southern European countries, with the notable exception of Portugal, scored even below the Belgian level. At the same time however, the Scandinavian welfare states with extremely high shares of social spending were also highly successful in employment terms, reaching levels that were as high or even slightly higher than the United States. Overall, it should be said, the statistical association between employment and social spending is practically zero. So much for conventional wisdom?



Exposed and Sheltered Sectors


But then how can one account for the fact that the most expensive welfare states with the highest tax burden among OECD countries and with powerful unions should be doing exactly as well in employment terms as the United States, which has practically ceased being a welfare state and has just about the lowest tax burden in the OECD, and where unions have lost all control over wage levels and structures? And why is it that Continental welfare states at similar levels of economic development and with intermediate levels of tax burdens should be doing so much less well?


In searching for an explanation, practically all contributions to the current debate tend to focus on international competitiveness. Thus one might expect the non-taxed, deregulated and de-unionized U.S. economy to have comparative advantages in sectors exposed to international competition, while the Scandinavian welfare states would achieve high levels of employment in the sheltered sectors. Unfortunately, this theoretically interesting distinction is not directly represented in the employment data available in the OECD Labour Force Statistics. Also, the boundary is shifting as hitherto sheltered jobs - for instance in telecommunications, financial services or in the construction industry - are becoming exposed to foreign competition with the completion of the European internal market and under the new WTO (World Trade Organization) rules. Opting for the most comprehensive definition, I have included employment in all ISIC (International Standard Industrial Classification of all Economic Activities) major divisions whose products are, actually or potentially, exported or subject to import competition. This definition includes not only all agricultural and industrial employment, but also service employment in ISIC Divisions 7 ("Transport, Storage, and Communication") and 8 ("Financing, Insurance, Real Estate, and Business Services"). Taking these branches together, and focusing now more narrowly on the United States and the countries of the European Union, Sweden and Germany, the outcome is truly surprising.



Again, the overall statistical effect of the size of the welfare state is extremely weak (and in fact slightly positive). Even more remarkable, the United States is doing rather poorly in the exposed sectors, whereas some of the Scandinavian countries and, among the Continental countries, Germany, Austria and, remarkably, Portugal, are doing much better. In other words, the size of the welfare state as such does not seem to have any negative effect on employment in the internationally exposed sectors of the economy.


The implications of these data for the present discussion are obviously ambivalent: On the one hand, countries with a high employment ratio in the exposed sectors will find a larger share of present jobs affected by the increasing pressures of international competition and by, perhaps temporary, downturns in the international demand for their products. This surely helps to explain the near-hysteria of the German Standortdebatte over the last few years.


On the other hand, one should not forget the more important part of the message: In the internationally exposed sectors of the economy, the "Rhenish model" of stakeholder-oriented corporate governance and cooperative industrial relations has been, and is still, doing very well indeed in international comparison. That does not mean that there is no need for reforms - on the contrary. But it does mean that there is no reason to throw the baby out with the bathwater, and to turn instead to American - or for that matter, British - models of market-driven corporate controls and industrial relations. In any case, however, the data demonstrate that the generally poor employment performance of, say, the German economy over the last decade cannot be ascribed to a general loss of international competitiveness. Instead, its explanation is to be found in the sheltered sectors of the economy.



Within the definition used here, these comprise the service branches in ISIC 6 ("Wholesale and Retail Trade, Restaurants and Hotels") and in ISIC 9 ("Community, Social, and Personal Services") - a heterogeneous collection which, however, shares the characteristic that local demand is being served by locally supplied services, and that foreign competition plays practically no role. It is in these "local services" that the data show a significant difference that could lead to an explanation of the poor employment performance of Continental welfare states.


Again, there is practically no linear relationship between employment and the size of the welfare state. Instead, the curvilinear relationship which exists (R[ 2] = .30) is U-shaped, with high employment at the upper and lower ends of the welfare-spending scale, and low employment in the Continental welfare states that are characterized by intermediate levels of welfare expenditures. Thus, in the United States altogether 41 percent of the working age population have jobs in the local services, and Sweden is not far behind at 39 percent. In Austria, Germany, France and Italy by contrast, the employment/population ratio of local services reaches only 28 percent - 13 percentage points less than in the United States (which would be equivalent to 6 million jobs in Germany). Figures in Denmark are almost as high as in Sweden, and Britain and the Netherlands are also doing better than Germany. In other EU member states, however, employment in the local services is even lower than it is in Germany. It is worthwhile therefore to further explore the underlying patterns.



Public and Private Services


In general, local services may be financed from either public or private sources, and they may also be produced by public agencies, by commercial firms or by not-for-profit organizations. As it turns out, it is this difference which finally points to an explanation. There is a weak (R[ 2] = -.10) negative association between welfare spending and local services provided in the private sector 1 - with the United States and the Netherlands as the upper outliers, whereas France and Belgium have very few jobs in private services.


At the same time there is a stronger (R[ 2] = .36) positive association between welfare spending and employment in the public sector - with Sweden and Denmark as the positive outliers. The Netherlands, but also Italy and Germany are here located well below the regression line.


Germany, for instance, is not doing much better than Sweden with regard to local service employment in the private sector, whereas - in spite of much higher tax levels, and contrary to a near-consensus in public debate - the German public sector is not at all "bloated". In fact it is just as "lean" in employment terms as is true of the United States.




The curvilinear pattern of employment in the sheltered sector as a whole is thus a composite result of two separate effects: In the United States and other countries with a small welfare state, local service jobs are created in the private sector of the economy, whereas in Denmark and Sweden with large welfare states, the public sector is able to provide high levels of local service employment. But why should the Continental welfare states with intermediate-size social expenditures have the worst of both worlds, instead of combining intermediate levels of employment in the public and in the private sector to achieve equally high overall levels of service employment?



Two Roads to Service Employment


The explanation, I suggest, lies in differences in the levels and structures of national welfare state and industrial relations systems. In order to simplify the argument, I will now concentrate on three models only, the American, the Swedish and the German, even though I know that there are significant differences among the Scandinavian welfare states and even greater ones among the Continental systems that need to be brought out by much more careful analyses than I could present here. In the American and Swedish cases, the explanations for their exceptionally high levels of employment in the sheltered service sectors are fairly straightforward.


In the United States, generally low levels of taxation and the additional tax cuts of the 1980s have contributed to a highly unequal distribution of incomes which has contributed to a simultaneous expansion of service employment at the upper and at the lower end of the skill scale. Since education and health care are to a much larger extent than elsewhere privately financed, the growing demand of affluent consumers for high-quality educational and medical services increases the number of well-paid jobs at the professional level.


At the same time, however, weak or nonexistent unions combined with the short duration of unemployment benefits and the virtual absence of social assistance for the long-term unemployed have facilitated the emergence of a low-wage labor market. This, in turn, has facilitated the creation, or maintenance, of service jobs in hotels, retail trade, restaurants, and a great variety of other household and personal services. In these jobs, labor productivity tends to be very low. But since wages are also very low, the American model allows large numbers of low-skilled workers to find employment in the private sector.


Thus, the up side of the American model is the dynamic expansion of service employment at all qualification levels. Its down side is the plight of the "working poor" receiving incomes below the subsistence level even when employed full time.


In Sweden and Denmark, by contrast, very high taxes, strong unions and generous rates of income replacement in case of unemployment have reduced income inequality and wage differentials to the lowest level among OECD countries. At the same time, education and health care are publicly financed and publicly provided. As a consequence, there is no low-wage labor market and little room for private services at the professional level. Instead, high levels of tax revenue are used to finance universal health services and education as well as a wide range of free social services for families with young children, for the elderly, the handicapped, 2 and the sick, for drug addicts and immigrants. These services involve not only work for highly trained professionals but also provide a large number of decently paid jobs for persons with relatively low levels of formal training. The down side of the Swedish model is, of course, its dependence on very high levels of taxation which, in the face of international tax competition for mobile capital and the growing tax resistance of mobile professionals, have become increasingly difficult to sustain. The result was, first, unsustainable budget deficits and, in the 1990s, a need for fiscal consolidation that forced governments to reduce not only the generosity of welfare payments, but also the level of public-sector employment - with the consequence that unemployment has risen to normal European levels, even though the level of employment continues to be very high by international standards.



The Continental Dilemma


By contrast, the continental failure to create high levels of employment in the local services cannot be simply explained by differences in the size of the welfare state. Total social spending amounts to only 15 percent of GDP in the United States, and to more than 37 percent in Sweden, whereas Continental welfare states tend to absorb between 25 and 30 percent of GDP (OECD 1996b). If nevertheless public-sector employment on the Continent is as low as it is in the United States, that appears as a path-dependent consequence of the Bismarckian model, that was originally meant to deal only with the social risks arising if the single (male) breadwinner was unable to support his family through full-time work (Esping-Andersen 1990). Hence, Continental welfare states are quite generous with regard to transfer payments in cases of retirement, disability, and unemployment - but they have never developed a Scandinavian commitment to provide social services that would complement or compete with the functions performed in the family by mothers, wives and daughters. As a consequence, the Continental welfare state certainly does not help to increase employment very much.


There are of course well-paid professional jobs in public education and in publicly-financed health care provided in public hospitals or in private practice. But the mere fact that these services are publicly financed, from tax revenues or from social insurance funds, means that employment growth is held back or even reversed by efforts to reduce tax burdens and public-sector budget deficits, whereas in the United States employment has expanded in response to the increasing demand of affluent consumers for high-quality services in education and health care.


At the low end of the labor market, however, the Continental welfare state is as effective as its Scandinavian counterpart in preventing the viability of a low-wage service jobs in the private sector. Compared to the United States, at any rate, levels of taxation are high, unions are strong, income inequality and wage differentials are low, and reservation wages above the subsistence level are assured by relatively generous and continuous social assistance payments. The negative impact on service employment is particularly acute in those countries which, like France and Germany, rely to a large extent on payroll taxes for the financing of the welfare state. In Germany, for instance, 74 percent of total social expenditures were financed through workers' and employers' contributions to social insurance systems in 1991, and in France that was true of 82 percent. In Germany these contributions presently amount to about 42 percent of the nominal wage paid by the employer.


In general, of course, there is no reason to think that high rates of payroll taxes (as distinguished from rate increases) should have a more negative impact on employment than other forms of tax finance. In the medium term, they will be taken into account, like all other factors affecting the total wage bill, when employers and unions are bargaining over wage increases - with the consequence that the "tax wedge" reduces the net income of the workers. At the lower end of the pay scale, however, things are very different. Here, the availability of social assistance defines the floor below which net wages cannot be reduced. Thus, certain types of "bad jobs" which are economically viable in the United States simply could not exist in Europe - which is, of course, a fully intended effect of the welfare state. What is not intended, however, is the impact of payroll taxes on jobs well above the subsistence level.


If the net wage of the worker cannot fall below a guaranteed minimum income, the consequence is that any social insurance contributions, payroll taxes and wage taxes that are levied on jobs at the lower end of the pay scale cannot be absorbed by the employee but must be added to the total labor cost borne by the employer. Assuming that any other overhead costs are proportional to labor cost, the implication is that the minimum productivity that a job must reach in order to be viable in the market is raised by more than 50 percent above the level of productivity required to pay the worker's net wage. As a consequence, a wide range of perfectly decent jobs which, in the absence of payroll taxes would be commercially viable, are driven from the private service market.


In Germany, to give a numerical example, social assistance for a single person amounts to about DM 1,100 per month, in addition to which that person may earn up to DM 250 by part-time work. Any additional income from work, however, is fully set off against social assistance - i.e., is taxed at a marginal rate of 100 percent. Assuming that an additional financial incentive is needed to make full-time work attractive, the minimum net reservation wage in Germany may presently amount to about DM 1,600 per month (i.e. about DM 10, or a bit less than US $ 6, per hour). In order to provide the worker with that net wage, however, the employer will have to pay a total wage bill of at least DM 2,400 per month which - if due allowance is made for other taxes and overhead - will allow only jobs with fairly high productivity to survive in the private sector.




In other words, financing the welfare state through social insurance contributions added to the wage bill has the effect of cutting off a fairly large segment of the labor market for jobs with low productivity. In this sense, therefore, it is indeed fair to say that Continental welfare states are causing high levels of long-term unemployment for persons with low levels of marketable skills.





The conclusion is that we have not two, but three, distinct models of service employment and welfare-state arrangements - American, Scandinavian and Continental-European. Both of the former are able to provide high levels of employment in the domestic service sectors - the United States in the private sector and the Scandinavian states in the public sector. Both also have their characteristic weaknesses. The downside of the American model is the plight of the working poor, whereas the Scandinavian model has become highly vulnerable as the capacity to tax is eroding for economic and political reasons. The Continental welfare states, by contrast, are not able to expand local services either in the public or in the private sector.


Moreover, while fairly straightforward solutions are available for the basic problems associated with either the American or the Scandinavian models, solutions for the weakness of Continental employment are less clear and, at any rate, much more difficult to implement. The easiest option exists in the United States, where large numbers of service jobs exist in the private sector, and where an appropriate solution for the poverty of large segments of the working population is already available in the form of the Earned Income Tax Credit program. There seems to be no reason why the goals of maintaining high levels of employment and of reducing mass poverty could not be realized in combination through an expansion and systematization of income support for the working poor.


In the Scandinavian welfare states, by contrast, large numbers of service jobs exist in the public sector. Their financial viability is in doubt, however, as the capacity of the state to tax mobile capital and the high incomes of mobile professionals is eroding. Nevertheless, if the commitment to high-quality public services for all is to be maintained, a straightforward solution might shift a large part of the financial cost of these services to means-tested user charges. If high-income groups are no longer willing to pay high taxes, they could at least be made to pay for the services which they are in fact using. Social justice could then be maintained by providing vouchers for low and medium income families - which would have the attractive side effect of introducing a strong element of consumer power into the governance of education and other service sectors.


The Continental situation is more difficult because the local service jobs that are needed to increase the overall level of employment do not yet exist either in the private or in the public sector. In the abstract, Continental Europe might consider moving either in the American or in the Scandinavian direction. If both options were equally feasible, political preferences would widely diverge. Practically speaking, however, the Scandinavian option appears to be out of the question. The reasons are financial and political. Continental welfare states, even though less expensive than their Scandinavian counterparts, are already hard-pressed financially. Thus, a Scandinavian-type service expansion would either require substantial tax increases or substantial reductions of welfare transfers. There is, however, no large and well organized political demand for additional services, whereas the political opposition against either tax increases or further cutbacks in social transfers is already highly mobilized. Thus a Scandinavian solution is not even discussed on the Left, whereas the "American way" is strongly advocated by business interests and the neoliberal opponents of the welfare state. It is equally strongly resisted by unions and political parties defending the "Rhenish model". As a consequence, change is blocked and mass unemployment continues to rise. What is needed is a re-framing of the public debate to overcome the present deadlock between the advocates of "work with poverty" and the defenders of "welfare without work". What is needed, in other words, are strategies that are able to combine "work with welfare".


To be effective and politically acceptable under Continental conditions, employment-increasing reforms would, at the same time, need to reduce the wage bill paid by the employer for low-productive private services and to assure that workers taking such jobs nevertheless would receive net incomes well above the subsistence level defined by present social assistance programs. In principle, both conditions could be achieved by a negative income tax - which is presently being discussed in Germany under the name of "Bürgergeld". It is illustrated for a single person without income from sources other than work.




Even if the reservation wages of workers at the low end of the labor market remained where they are presently assumed to lie, and even if all other rules were unchanged, the proposal would reduce the total wage bill of the employer by more than one third, from DM 2,400 to 1,500. It would thus open up a wide range of economically viable employment opportunities at the lower end of the labor market. There is every reason to think that the same solution would also work in all other Continental welfare states.


Politically, however, the negative income tax is a difficult solution. Combining income support for persons who are unable to work with income subsidies for low-income workers, it would require the simultaneous and far-reaching restructuring of the present systems of taxation, social assistance, social insurance pensions, and wage setting. Moreover, the available estimates of its overall financial consequences (in the absence of employment effects) are still widely diverging. Thus the chances are slim that German politics in its present shape could soon adopt this - theoretically optimal - solution to the Continental employment problem. Many of these difficulties could be avoided by a more modest and much simpler proposal which would not deal with the problems of income support for persons that are not working. It would simply provide income subsidies to workers in low-wage jobs below the present effective minimum, and leave all other present rules as they are.


In both versions, however, the employment effect of income subsidies would depend on the cooperation of unions that would have to agree to the creation of new wage scales below the present low-wage level. That is not an attractive function for unions that see their mission in raising, rather than lowering, minimum wages. Moreover, unions fear that wage reductions at the lower end could induce a general downward slide of the present wage structure. Even though that objection may not be theoretically plausible, union opposition has so far prevented political parties in Germany from responding to proposals of this nature.


Nevertheless, there is now a slight hope that a functionally equivalent solution - which would not depend on the active cooperation of the unions - might find more political support. Its feasibility rests on precisely those features of the Continental welfare states that are most damaging to service employment - namely their dependence on social insurance contributions from employers and workers as a major source of welfare finance. In Germany, I said, these amount to about 42 percent of the employer's wage bill, and they are shared equally between employers' and worker's contributions. Hence if these contributions were (almost) completely waived at and below the present effective lower end of the pay scale of about 10 DM per hour, the wage bill of the employer would be reduced by 22 percent - which would increase the profitability of service employment by the same percentage. At the same time, the take-home pay of the worker would also increase by the same percentage - which would increase the attractiveness of low-end jobs and, perhaps, would also allow jobs below the present minimum wage to be created.


Of course, these jobs would still need to be fully covered by the social insurance system. Hence the contributions waived would have to be made up by insurance payments financed from general tax revenues. Like the negative income tax, the size of the subsidy would have to decrease at higher wage levels. Thus, one might eliminate contributions almost totally at and below the level of the present minimum wage, and the subsidy could be reduced to zero at twice that level. These are obviously matters for political judgment and compromise, which would also have to vary from one country to another. But in principle, I suggest, this avenue should be open to all countries that are financing very large shares of their social expenditure through non-wage labor costs.


In conclusion, I have tried to show that the Continental employment deficit does not seem to result from a loss of international competitiveness. It affects the sheltered sector only; it is caused not by the size of the welfare state, but by its characteristic structure and mode of financing; and these causes could be remedied by institutional reforms that would increase, rather than destroy, the level of social-policy support for disadvantaged groups in our societies.


In other words, the rigidities that stand in the way of higher employment seem to be the rigidities of political systems that are incapable of effective reforms, rather than rigidities of the labor market.





Esping-Andersen, Gøsta 1990: The Three Worlds of Welfare Capitalism. Cambridge: Polity Press.
OECD 1996a: Labour Force Statistics. Paris: OECD.

OECD 1996b: Social Expenditure Statistics of OECD Member Countries. Provisional Version. Paris: OECD.


1   Again, these data are not directly available, but they can be obtained by deducting OECD figures on employment in "government services" from total employment in ISIC 6 plus ISIC 9. It should be noted, however, that the focus is on provision, rather than on finance. Thus services performed by private physicians or by charities would be included here even if they are financed by the state or by social insurance funds.

2   In 1993, Sweden spent 6.39 percent of GDP on "services for elderly and disabled people" and on "family services". In Denmark, the percentage was 4.36 and in Finland 3.01. The level is much lower in Luxembourg (1.23%), in the Netherlands (1.16 %), in France (1.11%) and in the UK (1.05%), and lower still in West Germany (0.74) and Ireland (0.53). Expenditure on these services is minimal in Belgium (0.36%), Italy (0.30%), Spain (0.20) and Portugal (0.13%) and practically nonexistent in Greece (OECD 1996b).


Copyright © 1997 Fritz W. Scharpf

No part of this publication may be reproduced or transmitted without permission in writing from the author.
Jegliche Vervielfältigung und Verbreitung, auch auszugsweise, bedarf der Zustimmung des Autors.

MPI für Gesellschaftsforschung, Paulstr. 3, 50676 Köln, Germany





MPIfG:  MPIfG Working Paper 97/7

[Zuletzt geändert am 29.03.2007 10:59]