16 J. C. RAY mean is being used here; this is calculated not over the whole population or a representative sample, but across the 84 typical cases of this study._______ 12 The problem of observations at the extreme (which can play an important and determining role in certain results) takes on a peculiar dimension here because of the nature of the data: as we are dealing here with multiple regressions, it is in principle the impact of each item of data on each regression coefficient which must be studied. What is under observation here is one case from one country; technically easy to carry out, an analysis of the effeccs of such cases would require long and, in this case, uninteresting commentaries, since here we are dealing with only particular cases and not a whole country. The question is whether these unusual cases should be suppressed in some countries and not in others. As for countries which are possibly atypical, it is not very easy to identify if the country is indeed unusual, since it is not just one element which must be scrutinized, but all of the cases for that country. It is therefore necessary to undo all of the regressions, taking each country in turn, then compare each of the regressions to the regression undertaken for the remaining countries together, and to do this for each regression coefficient. Systematic analysis of the results such as these, which should also be undertaken with weighting in order to take account of the respective populations of the different countries, exceeds the limits of a study such as this. 13 One variant of this type of analysis consists of applying weightings to each case-study according to the population of the country in question, diminishing the role of countries such as Luxembourg and giving a far greater role to a few large countries of the European Union. The results obtained would differ greatly from those which we have just presented, which is to be expected given the huge demographic differences among the European member states. 14 When the United States is included in the analysis, no weighting is added to the regressions because the difference in populations is so huge - 1 to 658 in the case of Luxembourg, and 1 to 4.4 in the case of France - as to render the results of no interest. 15 It is interesting, from a methodological point of view, to consider how far international comparisons are shaped by the choice of the country taken - implicitly or explicitly - as a point of reference. There are two points to be considered. First, as far as the definition of questions and categories is concerned, prior to the collection of data, it is right that where a particular national authority has commissioned the study, as in this case, the national context concerned should be given priority. Had it instead been funded by the Irish government, it would no doubt have given greater attention to large families; if by the French government, to various groups of low-income lone parents to whom French policies would have given specific attention. Second, as far as the analysis of the data is concerned, for example by means of multiple regression, the choice of one country as the point of reference is by no means neutral, since it can affect the impression which the reader is given as to the generosity or otherwise of (in this case) benefits in other countries. References Bradshaw, J., Ditch, J., Holmes, H. and Whiteford, P. (1993a) A comparative study of child support in fifteen countries, Journal of European Social Policy 3 (4): 255-71. Bradshaw, J., Ditch, J., Holmes, H. and Whiteford, P. (1993b) Support for Children: A Comparison of Arrangements in Fifteen Countries, DSS Research Report Series no 21. London: HMSO. Journal of European Social Policy 1997 7(1) Article SOCIAL ASSISTANCE IN OECD COUNTRIES JanJľk>ugh,4iniversity~ofi-Bathy-UK- Jonathan Bradshaw and John Ditch, University of York, UK Tony Eardley, University of New South Wales, Australia Peter Whiteford, Department of Social Security, Canberra, Australia Summary Résumé This article presents selected results from the first comparative study of social assistance across all 24 countries of the OECD. The scope of social assistance, discussed in the first section, is drawn to include all means-tested benefits in cash and kind, including those which provide benefits to higher income groups. The second section then presents information on the main programmes in each country, expenditures and groups of beneficiaries, trends over time, administrative structures, and operation of means tests. It concludes by developing a new measure of assistance benefit levels with which to evaluate different countries' systems. The third section distils from the country differences eight patterns, or 'assistance regimes', varying from the limited, discretionary, decentralized models of Switzerland and Norway to the extensive, national, rights-based programmes of the English-speaking world; and from the relative generosity of Scandinavia and Australia to the low, marginalizing benefits of the Mediterranean countries and the USA. The last section turns to the economic pressures and political debates which are driving contemporary policy changes. The concepts and empirical data presented here will enable means-testing, targeting and selectivity to be brought back into the comparative study of European and wider welfare systems. ĽAIDE SOCIALE DES PAY: ĽOCDE Cet article présente des résultats s. partir de la premiére etude cotnpc sociale dans les vingt-quatre pays Ľ aide sociale examinee dans section couvre toutes les prestatioi et en nature soumises ä enquéte su, y compris Celles qui sont fournies ä revenus plus élevés. La deuxi présente des informations sur le: programmes de chaque pays, les les groupes de bénéficiaires, les tet, le temps, les structures administ fonetionnement des enquétes sur Elles se conclutpar ľélboration ď, mesure des prestations ďaide : laquelle évaluer les systémes des di partir des differences entre les pays section extrait huit modeles, oi d'aide», qui vont des modéles ď ah isée, diserétionnaire et limitée de L la Norvége aux programmes tu grande envergure fondés sur le monde anglophone; et de la génén de la Scandinavie et de ľ Austrálie tions faibles et marginalisées des p ranéens et des Etats-Unis. La den se tourne vers les pressions éconoi débats politiques qui déterminem actuelle des politiques. Les con données empiriques présentés section pemtettront de ramener l les revenus, le ciblage et la sélt ľétude comparée des systémes d, sociale européens ou ď envergure / Journal of European Social Policy ISSN: 0958-9287 Crown Copyright © 1997. Published by permission of the ( Her Maiesty s Stationery Office. No part of this publication may be reproduced in any form without the nermiss Deparrment of Social Secunry. SAGF. Publications, London, Thousand Oaks and New Delhi, Vol 7 (I): 17-4 \ 18 IAN GOUGH ET AL. Introduction There is a mismatch between the salience of means-tested social assistance in practice and in academic study. In the English-speaking world' in particular, issues of targeting, 'welfare dependency', means-testing and behavioural incentives remain high on the political agenda. In many other OECD countries, both claimant numbers and costs of social assistance have been climbing because of rising unemployment and social and family change. Encouraged by the World Bank and other international agencies, selective 'safety nets' are widely recommended for the developing world and the new market economies of central and eastern Europe as the essential social protection measure to accompany economic reforms. Interest in targeting, selectivity and income-testing continues to grow, in the West (Room, 1990), the East (Cornia and Sipos, 1991) and the South (Burgess and Stern, 1991). Yet there has been remarkably little cross-national analysis of social assistance and other income-tested schemes even in the Western world.2 We have little comparative knowledge of the nature of different schemes, how benefits are calculated, whether they are rights-based or discretionary, the conditions attached to benefits, the levels of government responsible for financing and operating them, their effectiveness in reducing poverty or their efficiency in targeting those in need. Leibfried (1993: 139) has argued that concentrating on the margins of the welfare state is important because it is here that the limits and the content of social citizenship are best tested. This paper draws on recently completed research studying the full range of means-tested assistance programmes operating in the 24 OECD countries in the early 1990s. Data on social assistance arrangements were collected using two questionnaires: one sent to officials in the appropriate government departments, the other to academic experts. The replies to the questionnaires were collated into a single national report which was sent back to the informants for validation and a volume has been produced comparing the systems country by ^ountry-fE^rdley-et-alT^-LP^ébí^This-volume provided much of the raw data for another more analytical report in which arrangements were compared and contrasted across countries (Eardley et al., 1996a). This article is in four main sections. The first presents a typology of social assistance to order the variety which we uncover. The second section presents a comparative survey of key features of social assistance across OECD countries. This forms the basis for a 'map' of 'social assistance regimes' in the third section. The final section brings the story more up to date and surveys current debates on the future of assistance, interpreting these in terms of the common pressures faced by different assistance systems. The limits of this article should be noted. Our exercise is clearly 'Linnaean' rather than 'Darwinian': the aim is to chart and classify the species of social assistance we observe in the world rather than to offer a comprehensive theory to explain their variety and different forms of evolution. Nor do we evaluate means-testing and forms of social assistance according to independent evaluative criteria (cf. Gough, 1994).3 A typology of social assistance The term 'social assistance' does not have a fixed or universal meaning. In some countries it embraces a wide range of non-resource-tested but categorically targeted aid for such groups as orphans, immigrants and older people. In others (frequently the same countries) it excludes means-tested or income-related benefits administered as part of social insurance. There are some countries where assistance includes extensive services in kind as well as cash benefits. There are only three mechanisms by which the state can directly4 allocate income or services to individuals or households (Atkinson, 1989). The first is the 'universal' or contin- Journal of European Social Policy 1997 7(1) SOCIAL ASSISTANCE IN OECD COUNTRIES gency benefit allocated to all citizens within a -certain_sociaLcategoJ7JJrhese_henefits_are_no_L related to income or employment status. Second, there is social insurance, where the benefit is related to (a) employment status and (b) contributions paid. The third comprises means-tested or income-related benefits where eligibility is dependent upon current or recent resources, though other categorical conditions may also apply. Our study focuses principally on this third category of resource-tested benefits. Within resource-tested programmes we make three preliminary distinctions (Gough, 1994): 1. The first is between 'poverty-testing' and 'general means or income-testing'. The former provides resources to people who would otherwise fall below a certain, "Usually óffTčTálly~3ěfini3~ minimum living standard. This often reflects a political judgement rather than a scientific assessment (Veit-Wilson, 1993). It may or may not be referred to as a poverty standard, but there is some recognition j)f providing a 'safety neť _below_which nobody should fall. General means-testing, on~the^other hand, is concerned to relate benefits to current resources across a broader range of income groups - it may be no more than a means to restrict access by the well-off (cf. Eurostat, 1993: 6). 2. There is also a distinction between cash, and „'tied' benefits. The former provide money benefits for either exceptional needs or regular payments. Tied' benefits entitle recipients to free or subsidized use of a specific service or to a refund < of the charge for a specific ser ticularly important tied bene countries is housing assistance. 3. There is an important di: tween schemes for all people ' tain income/resources group awarded to more specific catei this group, such as older or < sons. These three distinctions generate nations as shown in Figure 1. One approach would be to ig eral income-related benefits in ce as the growing numbers of incom ily allowances. However, this cat in Australia and New Zealand w social insurance features. Their s systems are, in the terms introi almost entirely resource-tested, b cused solely on those with minin To limit our study to 'povert) grammes would be to exclude vii tipodean benefits except for Special Benefit. Moreover, wheth efits extend to those with higher purpose of such schemes is still t minimum level of subsistence < fined) to those at the bottom. However, to be consistent we include Family Credit in Brit related family allowances in Italy the Earned Income Tax Credit Arbeitslosenhilfe in Germany, lowances in Sweden, and so on. exclude all tied benefits or bei from our study would be to hir All groups Specifii Poverty-tested Cash 1 2 Tied 3 4 General means-tested Cash 5 6 Tied 7 8 Figure 1 Taxonomy of means-tested schemes Journal of European Social 20 IAN GOUGH ET AL. son with countries where these are important, for example Food Stamps in the USA. -Moreover^-the degree ofintegrationof-housing benefit with social assistance varies across countries, so to exclude it would distort comparisons. This paper therefore embraces all categories of means or income-tested benefits in Figure 1, whether or not they are regarded as assistance within the countries themselves. A comparative map of social assistance A tally of national programmes Table 1 allocates all major means-tested programmes in OECD countries to one of four categories: general, categorical, housing assistance, and other tied assistance. The nature of schemes can change over time and this table (and this article) relate to the situation in May 1992. It excludes separate schemes for one-off and emergency payments. In the OECD, only Portugal and_Greece lack any general, non-categorical assistance programme, though Turkey's is extremely limited and those in Spain and Italy. vary_ markedly between regions. Most also have one or more group-specific assistance schemes. These are especially important in Australia, New Zealand, the USA, Belgium, France, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Portugal and Spain. Th^JJSA is unique in the range of other tied assistance programmes^ The majority oftííěse programmes are targeted at low-income households, families or individuals, but in several countries there are income-tested programmes (of which family allowances are most common) which extend their benefits further up the income scale, through extensive earnings disregards or tapered benefit withdrawal rates (indicated by an asterisk in the table)/Australia and New Zealand are unique in the extent of such schemes?) The salience of social assistance The^salienee of-soeial-assistanee ean-be^eharted by the size of state expenditures and the proportion of beneficiaries in national populations. Table 2 shows, in the first column, the total estimated numbers of beneficiaries in households receiving any form of social assistance (not the number of households) as a proportion of the total population.6 Columns 2 and 3 present estimates of total assistance expenditure as a share of GDP and social security expenditure. Given the difficulty in obtaining a consistent definition of social security expenditure across countries, we place greater reliance on the GDP share indicator. The table reveals considerable variation in the importance of social assistance in different countries. The proportion of the population receiving social assistance varies from 0.7 per cent in Japan and Greece to 25 per cent in New Zealand, 18 per cent in Australia and 16 per cent in the UK. Total assistance expenditure varies between 0.1 per cent of GDP in Greece and 13.0 per cent in New Zealand; and as a share of social security between 1 per cent in Luxembourg and 100 per cent in New Zealand. Even excluding the exceptional cases of New Zealand and Australia, these are wider variations than for spending on other social programmes, such as social insurance. The salience of means and income-tested benefits is particularly high in a minority of countries. Whether we take more than 10 per cent of the population receiving benefits, or expenditure exceeding 3 per cent of GDP or 30 per cent of social security, the same five countries emerge on top: Australia and New Zealand, the UK and Ireland, and the USA. (Italy spends more than 3 per cent of GDP, and Iceland might also be in this group if spending on its pension supplements were included. Y, The social security systems of the English-speaking world clearly rely more on means-testing than elsewhe£ej>At the other extreme the countries are more mixed: Japan, the Benelux countries, Iceland, and Greece, Portugal and Turkey among the southern Journal of European Social Policy 1997 7(1) SOCIAL ASSISTANCE IN OECD COUNTRIES lil'' 111 Journal of European Social Poli Table 1 A taxonomy of national resource-tested programmes (1992)a (continued) Country General assistance Group assistance Housing assistance* Other tied assistance Italy Local cash assistance (Minimum Vitale) Japan Livelihood Aid (integral with Education Aid, Housing Aid, Medical Aid, Maternity Aid, Occupational Aid, and Funeral Aid) Luxembourg Revenu Minimum Garanti; AVC Netherlands Algemene Bijstand (ABW) New Zealand Special Benefit Norway Social Economic Assistance Portugal Spain Ingreso Minimo de Insercion Pensione Sociale; Pensione di Inabilita; Veteran's Pension; Assegno al Nucleo Familiäre* Children's Allowance* Allocation pour Personnes Gravement Handicappés; Allocation de Soins Regulation for Unemployed Employees (RWW); Income Provisions for Older and Partially Disabled Workers (IOAW); Income Provisions for Formerly Self-employed (IOAZ); Supplementary Benefit (TW) Unemployment Benefit*; Training Benefit* Sickness Benefit*; Invalidity Benefit*; Widows Benefit*; Domestic Purposes Benefit*; Transitional Retirement Benefit*; New Zealand Superannuation*; Family Support* Transitional Allowance Family Allowance; Supplementary Allowance; Nursing Allowance; Orphan's Pension; Social Invalidity Pension; Social Old Age Pension; Young People's Integration Benefit; Survivors' Grant Means-tested Age Pension; Disability Pension; Unemployment Assistance Local assistance services Allocation de Chauffage Housing Benefit Accommodation Supplement (from 1994) State Housebank Housing Benefit; Pensioners' Housing Benefit Housing Benefit ?.«,, .iwv.jnyEfl tm'WWiiimigSIWFW'W flPlfP^BWPf^^ > O o c o M H > r ^SWWWffP Table 1 A taxonomy of national resource-tested programmes (1992)a (continued) Country General assistance Group assistance Housing assistanceh Other tied assistance Sweden Socialhjälp Switzerland Soziale Fürsorge Turkey Social Assistance and Solidarity Scheme UK ) Income Support USA Food Stamps; General Assistance Supplementary retirement and invalidity pensions Old age and disability assistance Family Credit; Disability Working Allowance Aid to Families with Dependent Children; Supplemental Security Income; Veterans' Pension; Earned Income Tax Credit* KBT-Municipal housing allowance tor pensioners *; SKBT-State/municipal housing benefit* Housing Benefit Federal Housing Assistance; Low Income Home Energy Assistance Green Card medical assistance Community Charge Benefit (later replaced by Council Tax Benefit); Free School Meals Medicaid; School Lunch and Breakfast Program; Special Supplementary Food Program; Job Training Partnership Act; Head Start Notes * 'denotes income-tested scheme extending benefits to those with incomes above poverty level b That is, separate housing assistance programme, excluding those integrated into other assistance schemes. O Q > O w n o n o c z H S 24 IAN GOUGH ET AL. Table 2 > Indicators of the salience of social assistance Country SA recipients as SA expenditure SA expenditure as % of % of population as % of GDP sociaLsecutity expenditure----- ÍW2 1992 1992 Australia 17.8 6.8 90.3 Austria 4.8 1.3 6.7 Belgium 3.6 0.7 3.0 Canada 15.1 2.5 18.9 Denmark 8.3 1.4 7.8 Finland 9.2 0.4 n/a France 2.3 1.8 6.4 i Germany 6.8 1.6 ll;9a Greece 0.7 0.1 n/a Iceland 3.7 0.2 1.2 Ireland 12.4 5.1 41.2b Italy n/a 1.5 9.1 Japan 0.7 0.3 3.7 Luxembourg 2.7 0.4 1.4 Netherlands n/a 2.2 10.9 New Zealand 25.0 13.0 100.0 Norway 4.0 0.7 4.8 Portugal 2.1 ! 0.4 3.8 Spain 2.7 1.2 8.4 \ Sweden 6.8 0.5 6.7 Switzerland 2.3 0.8 1.8 Turkey n/a 0.5 n/a ' UK 15.3 6.4 33.0 ( USA 17.5 1.3 39.8 Notes: Column (3): Social security denominators from OECD Household Transfer Data Base, except for Australia, Belgium, Iceland, Luxembourg, Portugal, UK - where government calculations of social security used. These different denominators can affect the shares shown; for example, the shares for Germany are considerably higher than those shown in German government sources. ■ 1990 b1991. European countries all exhibit low levels of spending and low numbers of recipients. In between are found the majority of members of the (expanded) EU plus Norway. Patterns of social assistance Aggregate measures of this sort may hide as much as they reveal. Beginning with the numbers of individual beneficiaries of general assistance schemes, the unique scale of Income Support in the UK is striking. It delivers benefits to 15_p_er. .cent of the British population, a higher proportion than any other single programme in the OECD. The US Food Stamps programme, the various provincial benefits provided under Canada Assistance Plan, and in recent years the Finnish Social Assistance Allowance deliver to about one in ten of their populations; whilst social assistance in Sweden and Germany cover roughly 5 per cent. When categorical, group-specific schemes are included, New Zealand, Australia and Journal of European Social Policy 1997 7 (1) SOCIAL ASSISTANCE IN OECD COUNTRIES Table 3 Social assistance expenditure as a percentage of Gross Domestic Product (GDP), 199 Country Group assistance Cash Housing assistance Cash + Other tied T assistance assistance housing assistance sc [1] m [1+2] [31 assistance [1+2+3] [4] as [1 Australia 0.1 6.1 6.8 _ 6.8 - f Austria 0.1 0.7 0.8 n/a 0.8 0.5 1 Belgium 0.1 0.6 0.7 - 0.7 - "1 Canada 1.8 - 2.5 n/a 2.5 - í Denmark 1.4 n/a 1.4 n/a 1.4 - 1 Finland 0.4 - 0.4 n/a 0.4 - C France 0.2 1.0C 1.2 0.8 2.0 - 2 Germany ,0.5 0.3 0.8 „ 0.2 l-L- 0.9 2 Greece* - 0.1 0.1 n/a 0.1 - 0 Iceland 0.1 0.1 0.2 n/a 0.2 - 0 Ireland 0.3 4.8 5.1 - 5.1 - 5 Italy 0.2 2.7 2.9 - 2.9 0.4 3 Japan 0.2 0.1 0.3 - - 0.3 - 0 Luxembourg 0.4 0.1 0.5 n/a 0.5 - 0 Netherlands 0.8 1.4 2.2 n/a 2.2 - 2 New Zealand 0.1 13.0 13.0 - 13.0 - 13 Norway 0.5 0.2 0.7 0.2 0.9 - 0 Portugal n/a n/a 0.4 n/a 0.4 - 0 Spain 0.03 1.0C 1.1 - 1.1 - 1 Sweden 0.5 - 0.5 1.0 1.5 - 1 Switzerland 0.2 0.6 0.8 - 0.8 - 0 Turkeyb 0.5 n/a 0.5 - 0.5 n/a 0 UK 2.5 0.1 2.6 1.2 3.9 0.3 4. USA 0.4 0.9 1.3 0.3 1.6 2.1 3. Notes Expenditures refer to programmes shown in Table 1 unless otherwise stated. Numbers may not add up due to rounding. - indicates that there are no substantial forms of expenditure within the particular category, whereas : indicates that information is not available. »1988 b1993 c Excluding means-tested age pension (for which expenditure figures not available). Iceland predominate - fully one-quarter of New Zealanders, one-fifth of Icelanders and one in six Australians receive means-tested benefits.7 Ireland also joins Britain in the high-coverage countries. Aid to Families with Dependent Children (AFDC) and Supplemental Security Income (SSI) cover 7.5 per cent of the US population, but these overlap with the recipients of Food Stamps (though not entirely, which underestimates the total number in the USA). In the next group are Austria (mainly based on supplementär sion recipients), Canada, Finland and S^ At the other extreme, only 2 per cent oi of the population receive social assista France, Greece, Japan, Portugal, Switzerland. France, the UK and the Netherlands ; ceptional in the coverage of their housi sistance schemes. In Britain, beneficiarie: lap with those of Income Support to a extent, but in France they far outnumt Journal of European Social Policy 19 26 IAN GOUGH ET AL. recipients of RMI or the other assistance ichemesT-substantia-Hy—r-aisi-ng—t-he-profile-of— means-testing. Evidence on other tied benefits is sketchy, but two countries stand out: the USA with over 11 per cent receiving Medicaid, the most extensive assistance programme in the country; and the UK with a similar share receiving Community Charge Benefit in 1992. When housing assistance and social assistance in special circumstances is included, Germany also joins those countries with extensive populations receiving means-tested benefits. The pattern is similar when we turn to expenditure (Table 3). Looking first at general, non-categorical assistance programmes, the UK stands out - spending 2.5 per cent of GDP on Income Support - followed by Canada and Denmark. When categorical schemes are considered, the distinctive patterns of New Zealand and Australia but also Ireland are clearly revealed. Britain also leads the OECD world in spending on specific means-tested housing benefit (which goes to people other than just those on incomes at the assistance level), followed by two very different countries: France and Sweden. Other tied benefits are prominent in three countries. On the one hand there is the USA, notably as a result of its unique medical assistance programme Medicaid. On the other hand, Germany and Austria devote the bulk of assistance to residents of homes for older people and on payment for other care services. This may be a feature of social assistance expenditure in other countries which is not always apparent from official statistics. In most continental EU member countries, categorical assistance for specific groups is more important than general programmes. In particular, older and disabled people are separately catered for. This is the case in France (also for lone parents and widows), Belgium, Portugal, Italy and Spain. In Germany, Ireland, the Netherlands and Spain there are separate assistance schemes for unemployed people. In Europe outside the British Isles, categorical social assistance dominates in Belgium, France, Greece, Italy, the Jbfetherlands, Spain, Austria and Switzerland. Trends in social assistance iThe general pattern of social security in the '20th century has been for older forms of public assistance to be replaced by social insurance and, in certain countries, universal benefits. ;But in the 1980s a reversal of this process was ! noted in some nations. Atkinson (1989: ! 107-12) found a rapid decline in the share of ; British social security expenditure devoted to ! public assistance before the First World War. The new income-tested schemes, beginning with the 1908 Old Age Pensions Act, accounted for one-third of social security spending in 1914 and about one-quarter in the inter-war period. By 1949, the former had disappeared and the latter had reached a nadir of around one-eighth of a much enhanced social security bill. Means-tested benefit expenditure remained at this share for the next two decades, then began to rise again after 1979. Room (1990), writing on the emergence of a 'new poverty' in the EC, noted a similar trend for the numbers relying on social assistance to rise in the early 1980s, due above all to rising unemployment. We can now update and extend this work to cover all OECD countries from 1980 onwards. Table 4 shows trends in numbers receiving, and government spending on, 'cash' social assistance - that is, excluding housing and other tied assistance.8 The absolute changes in recipients range from the UK at one extreme (a growth in share of population of nearly seven percentage points) to New Zealand at the other (a decline of over five percentage points). Most countries record a substantial expansion in the proportion of beneficiaries, notably Britain, Australia, Canada, Ireland, Germany and the Nordic countries. The proportionate increase has been greatest in Scandinavia (except Sweden), France and Portugal. The expansion of cover- Journal of European Social Policy 1997 7(1) SOCIAL ASSISTANCE IN OECD COUNTRIES Table 4 Changes in cash social assistance (i.e. excluding all tied assistance), 1980-92" Country Recipients % Recipients % Expenditure % Expend: pop: Pop: GDP: GDP: Change Index <" Change Index 1980-92 1992/1980 1980-92 1992/1S Australia13 4.2 131 1.4 126 Austria* -0.4 -22__ 0.3 124. Belgium 1.7 189 0.2 156 Canada 4.1 170 0.9 197 Denmark n/a n/a n/a n/a Finland 5.7 265 0.3 438 Franced 1.5 296 > 0.2 (0.7) 196 (20 Germany1 2.9 1Z2_ - 0.6 160 Greece n/a n/a 0.0 100 Iceland n/a n/a n/a n/a Ireland 3.7 142 2.2 174 Italy 1.2 135 - 0.4 135 Japan -0.5 59 -0.1 60 Luxembourg n/a n/a' n/a n/a Netherlands 1.1 143 0.5 133 New Zealand5 -5.7 82 4.4 151 Norway 2.5 271 / 0.5 486 Portugal 1.2 241 0.2 221 Spain n/a n/a 1.0 473 Sweden 2.7 164 < 0.3 (0.7) 272(1* Switzerland' 0.5 130 y -0.1 89 Turkey n/a n/a n/a n/a UK 6.7 177 - 1.2(2.1) 190 (2' USA* 1.0 116 ~=r 0.2 115 Notes: Numbers in brackets include housing assistance for France, Sweden and UK. ■Increase to 1991 or 1990 where no data for 1992 b Social assistance data for financial years (1/7 to 30/6) related to later calendar year; e.g. 1991/92 as proportion of 1991 c Includes payments to residents in homes d API and AAH only 'Social assistance data for financial years. These ran from 1/4-31/3 until 1990, then from 1/7-30/6. assistance data related to earlier year until 1990 (e.g. 1988/89 as proportion of 1988) and to later ye: 1990 (e.g. 1991/92 as proportion of 1992). 'Supplementary pensions only 8 Recipients: SSI plus AFDC only. age of the major schemes in the USA (with the exception of EITC) is one of the lowest in the OECD. Overall, country differences in the coverage of social assistance narrowed over the'period. In several there was a marked expansion in the early 1990s. Four countries recorded an increase in ex- Joumal of European Social Policy penditure since 1980 of over one pei point of GDP: New Zealand, A Ireland and the UK (over two point housing assistance is included). Jap Switzerland are the only countries fo we have reliable data which register : ing share. Proportionately, again ii 28 IAN GOUGH ET AL. Nordic countries which have witnessed the most rapid growth, albeit from a low base. ^hey^arrfoBöweabythe UKand-Erance-(both-exacerbated by fast-growing housing assistance) and Portugal. The result is some slight narrowing of the gap in national expenditure on cash means-tested benefits. Though not shown in the table, the share of the social security budget spent on all assistance-type benefits also records an increase in every country except Japan and Switzerland. The fastest growth is still in some Nordic countries. Otherwise the proportionate increase is remarkably similar across all OECD countries. All types of welfare regime exhibited a rising share of expenditure on means-tested schemes in the 1980s - a notable convergence of otherwise disparate national patterns. Major groups of recipients It is difficult to obtain full, comparable data on categories of assistance recipients, but some information has been gathered for the study. * Older people constitute a major clientele in Australia and New Zealand, where there is no | social insurance, but also in the UK and i Ireland, where assistance benefits supplement relatively low basic pensions. In most other countries they are a small and declining group dependent on means-testing - the result of widespread improvements in social insurance. The numbers of disabled recipients (and their families) have expanded in all countries for which we have data and now constitute a significant group of recipients in Italy, Belgium, Iceland, Australia, Canada and the USA. is Lone parents form a large group of assistance beneficiaries in the USA, Canada, the UK and New Zealand, reflecting in part their incidence in the populations, and are significant too in Australia and Finland. Elsewhere they make few demands on social assistance, either because there are relatively few of them, because they are more likely to be working or be- cause they are better catered for through other non-means-tested forms of support. However, __the proportion of assistance beneficiaries who are lone parents has risen in all countries for which we have information, except for Japan and the Netherlands. There is surprisingly little comparable information on the proportion of beneficiaries who fl are unemployed, but for the majority of countries where this information is available they are the most numerous category and the fastest-growing since 1980. By 1992, unemployment was the cause of significant demand for means-tested benefits across the OECD, but especially so in the English-speaking countries except the USA. It has pushed another 4 per cent of New Zealanders and Canadians onto social assistance over the last decade. The impact of unemployment on social assistance claims is mediated by country differences in entitlements, the range and generosity of alternative benefits, and changes in the definitions of unemployment. f Several authors contend that women are I more dependent upon social assistance than ■ men. It is argued that welfare states are gendered in their structural separation of insurance-based, labour market ('masculine') provision, and family-based and means-tested ('feminine') social assistance, which leads to a distinction between , 'rights-bearing beneficiaries' and 'dependent clients' (Fräser, 1987; Sainsbury, 1993). We do not have the necessary data for all countries and the evidence is conflicting/in Germany, 62 per cent of all claimants are women and the proportion exceeds 55 per pent in many other EU member states. However, in the Nordic countries, the share is slightly below a half, owing to the extensive array of alternative benefits for women and mothers and the better provision for women in old age. The extent to which national assistance programmes are gendered clearly reflects the options for income support available to women in different welfare states. Journal of European Social Policy 1997 7 (1) SOCIAL ASSISTANCE IN OECD COUNTRIES The administration of social assistance There is not the space here to present more than a fraction of the information collected on the structure, administration and delivery of assistance programmes. Table 5 summarizes information on five aspects of administration: the degree of centralization, the extent of family obligations, the toughness of means-testing, and the extent to which benefits are a legal entitlement. The first contrast is between schemes which are uniform and those which show sub-national variations. To begin, we must demarcate federal states, where state or provincial governments have a constitutional role, from unitary states. In the USA and Canada there is considerable inter-state variation in benefits and eligibility, and in the Swiss confederation responsibility is located at cantonal and even at commune level (of which there are about 3000). However, in Australia there is virtually no variation and in Germany the Länder set benefit rates within a narrow band fixed nationally. Among the unitary states, there is a contrast between the English-speaking countries and the rest. All the former have national or national-state legal and administrative frameworks. At the other extreme, in southwestern Europe (excluding Greece), Sweden, Norway, Iceland, Austria and Switzerland, local governments have considerable power, though there remain differences between countries in the role of central government. Finland has been phasing in a scheme of national scales but there are still variations between communes. In Sweden there are nationally recommended standard scale rates, though in practice there is much variation at commune level. Norway has no national scale rates - if scales exist they are set and interpreted locally at commune level. Spanish regions set their own social assistance levels while those in Italy are permitted but not required to establish general frameworks for social assistance, and the extent to which they do varies markedly. In Austria provinces set benefit rates but district authorities are free to vary them. The most common model national guidelines administered National scales which are nationall) tered occur in the UK, the USA Stamps, New Zealand, Australia. Ireland, France and, with minor regi ations, Japan. A second question concerns fai gations. The majority of countries/t benefit unit the nuclear family rathe household. Only Austria and Sv Luxembourg, Japan, Turkey and th Food Stamps treat the household as unit. At the other extreme, Ice Denmark are unique in treating coh; separate units. Most countries alsc nuclear family as the resource un whose resources must be taken int when applying the means-test. The < here are Germany, Austria, Sv Luxembourg, Turkey and the USA Stamps. All these countries can in cumstances expect members of the other than the nuclear family and fa bers living elsewhere to contribute come of a claimant, if their income v The nature of the means-test is a t ture of social assistance regimes. O ant aspect is the extent to which e: disregarded. Table 5 (column 3) di several categories. First are those with relatively high earnings di; Australia, New Zealand^ most prog the USA," together with Belgium and Denmark ~couTčf perhaps" be incl though its high guideline disregai rarely to be applied. At the other e those countries with no earnings di "all - the Nordic coiintřiěš7~šSuthe Turkey, Austmand Switzerland. Ö aní'the' United Kingdom have rul excluding those working more than number of hours per week. Indee countries social assistance acts prii 'top-up' to part-time or low earnin In most countries, there is a usually requiring that recipients reg employed and establish that they . Journal of European Social Po 30 IAN GOUGH ET AL. r<1 (N ro r*l M ~ "í N N N n n l. =5 -c Er Er =* .S > S S "3 e 13 -2 .2 g -2 o< lil ** tli I, ^IIIIilľílJÍÍ!!íf!lJ^ nl nl n) cd SS-Ä Hlu ««'HSSJsbS1* «5 m OJD C C "^ 2 l-i u .—i 3 3 -b S « .g 2 ,U ,í S 5 e — S O 5 ai _ d .ä OD M •sb «- "2 u C a, J> ■a v -a ■a 3 -■' S fS ro <-*-! T8 ■g ^ H «u c n o o 111 •a ja _ -3 n "5b ví O, •-* ^ *S ■*"* n _ rt ?! P u u w 5i B" c h -S, &"' ä .Sř « T3 o ^ > .3 S i^-a ja Pi °" S P ■ 2 -p in 8 ti ! fit! i« ta Jj rt d rt ?J tu 1?— o 2 o 3 o B o— u — -,= -=• -=• w-= w -= -.h --íj ^ «i vn'TľJí—« c/ic-'c^cr* w.is rt.ü rt^C^^C--- rt"-> ctí rt ta ca ,,-< Journal of European Social Policy 1997 7(1) Social Assistance in OECD Countries looking for work. However, in virtually all čountŕíišT work-tests are notapplied, or are more relaxed, for people who are ill, experience disabilities, or who are over or approaching retirement age. The major variations relate "* to IonTpäré^ntsTTŕľparticular to the age of chil-"ciren who exempt lone parents from the requirement to actively seek work (Bradshaw et al., 1996). Virtually all social assistance schemes operate income-tests with a 100 per cent with-drawal rate above the assistance_ threshold. Exceptions are Australia, New Zealand^ Ireland for lone parents since July 1994, Portugal for social pensioners, and the United States. In the USA, lower withdrawal rates operate on earnings for Supplementary Security Income (SSI) and Aid to Families with Dependent Children (AFDC), and Food Stamps are reduced by only 30 per cent of increases in net income. Arrangements in Australia and New Zealand are the most liberal, reflecting the absence of social insurance benefits. Another important facet of the means-test is how assets are treated. Table 5 separates nations where most assets are counted and those where there are relatively high disregards on assets, especially dwellings. Here the patterns are surprising. A less strict pattern of treatment is found in the extensive social assistance regimes of the English-speaking world, with the exception of US programmes -particularly AFDC. Yet greater leeway on savings and owner-occupied homes is also common in most countries of the EU, including the limited assistance regimes of the Mediterranean world - where less developed administrative capacity may be an important explanation., It is in Scandinavia, Japan, Austria and Switzerland that the toughest tests on assets are found. Finally, there are two related features - the degree of discretion in the award of benefits and the extent of effective rights of appeal. At one extreme are the English-speaking countries with established appeals mechanisms, where official discretion is also minimized. In Finland, Sweden, Luxembou Netherlands and Germany applicant? similar degree of codified rights. At i extreme are countries where individ cials and/or local municipalities en siderable discretion in awarding benei of the Mediterranean countries, Swi Austria, and Norway. Social assistance benefit levels We now turn to the impact of these v on the final outcomes - the level oi provided to recmients. To^estimate have adopted (EhtPmodel families sii method - asking ha^gnairíňromänfs late the in^ac^ofsMiaJ^aj^istance:b( a selection of model families in spec cumstances.9 —— The method involves calculating, a point in time (May 1992) the net d resources of a given variety of family three situations - íeegivíňg^ôcíaTa ŕecČTvinípTSbcial insurance íuiď ^woi national average earnings. Account is income tax ändsociaí insurance conti and of any cash benefits received. Ass are made about housing costs, loc charges for medical treatment, the schooling and pre-school child c method provides comparable inforn an up-to-date basis, but there are se advantages. First, it produces a^ dese the way the system should work ra how it necessarily does work>Sec more assumptions made aboutth< stances of the model families the less tative are those families of real po| Third, in countries where assistance vary from municipality to municip; difficult to talk about a national soc ance scheme.10 We have calculated both absolute, ized and relative benefit levels i method. The first uses OECD p power parities (PPPs) to convert nat Journal of European Social Polit 32 IAN GOUGH ET AL. rency amounts into a common monetary denominator. PPPs are generally more satisfactory than exchange rates-beeause they take-account of differences in the price of a standard basket of goods and services in each country. Social assistance entitlements for families of different types vary considerably, so it is not easy to make overall comparisons of-the relative-leveLof-income_of recipients of social assistance in different countries. The first two columns of Table 6 show absolute values by taking the mean of the amounts in Table 6 Absolute, standardized and relative social assistance benefit levels, 1992 Country 1. Absolute 2. Absolute 3. Standardized 4. Standardized / Í. Relative ' ^ benefit levels: benefit levels: benefit levels: benefit levels: S benefit levels: before housing after bousing before housing after housing ^ after housing/ costs costs costs costs V. costs y Australia 15) 28 13 ' 25'' "5T""" Austria -29 -2 -35 -10 (74- Belgium 1 -12 -8 -20 v40' Canada 26 2 6 -14 35 Denmark ,26 29 18 20 '': 66 Finland -5 31 8 49, 78 France -28 • 4 30 -7 43 Germany -32 •- -5 -45 -24 r**> Greece -91 -119 -82 -113 -16 Iceland 49 50 44 45 0 Ireland -15 2 10 '32! 49 Italy (0) (28) (-5) (21) (60) Japan 1 8 15 -10 45 Luxembourg 32 35 0 2 43 Netherlands 16 33 13 '30 ■J$ New Zealand -4 -18 11 -6 43 Norway 38 25 29 17-' 57 Portugal -63 -90 -38 -83 19, Spain -45 -41 -29 -24 25 Sweden 17 24 17 24- .77) Switzerland 91 41 41 4 86 Turkey n/a n/a n/a n/a .0 UK -19 '■ 11 -18 13 (42 ) USA: NY 8 6 -24 -25 IT" USA: Penn. -24 ■ -23 -46 -46 29 USA: Florida -27 ' -64 -46 -75 27 USA: Texas -40 ' -85 -58 -89 6 Notes: Column (1): Disposable income of social assistance recipients (before housing costs) in purchasing power parities expressed as a proportion of the mean for all OECD countries: average for nine household types Column (2): As for column 1, but calculated after housing costs Column (3): Column 1 standardized for GDP per capita in 1992 Column (4): Column 2 standardized for GDP per capita in 1992 Column (5): Disposable incomes of social assistance recipients as percentage of disposable incomes of the same household types where the head is earning average male earnings: average of six household types (as above but excluding single pensioner, pensioner couple, and couple with one child aged 3) after housing costs. The figures for Italy may be unreliable. Source: Eardley et al. (1996a), revised. Journal of European Social Policy 1997 7 (1) SOCIAL ASSISTANCE IN OECD COUNTRIES purchasing power parity £ sterling paid to nine lamily "typesy'^lind expressing this as a proportion of the mean lor all TJECD countriesTtf Given the scale of the USA and the extent ol inter-state variations within it, we have calculated separate benefit levels for four states -New York, Pennsylvania, Florida and Texas. These were chosen as four populous states varying from among the most to the least generous. It must also be noted that where there are substantial local or regional variations in benefit levels, these calculations only apply to the particular location chosen for the study. This particularly affects Austria, Switzerland, Norway, Spain and Italy, the other Nordic countries and to a lesser extent Canada. Columns 1 and 2 show the results both before and after taking housing costs into-account. Housing costs present particular difficulties in comparative research. Costs vary both within and between countries according to tenure, the size, age and location of dwellings. In some countries rents may be controlled for persons occupying dwellings before a certain date. For owner occupiers the stage in their life cycle will tend to determine the level of the mortgage payable and interest rates vary between countries. Nevertheless, housing costs are too important to be ignored. In many countries help with housing costs is a critical element in the social assistance package and, even where such support does not exist, variations in housing costs mean that the real value of social assistance varies substantially. To balance the greater validity of the after-housing cost index against the greater reliability of the before-housing cost index, both arej?resented in Table_6.--------^_______ Top öT~the table, after housing costs come Switzerland, the Nordic countries, Luxembourg, the Netherlands, Australia and Italy (though there are considerable doubts about the Italian figures). At the bottom are the USA states, except New York, and the :ountries of southern Europe. In between are ound the remaining English-speaking counties, Japan and the north European members af the EU. There are a considerable number of changes in xanking Lciure and after costs - after housing costs Canada, and New Zealand move down the and Britain, Finland, Germany, Italy, and France improve their positions. Ab^lute_„YjJags_of_course_ faypu countriest_._Tp correct for this the n columns adjustjhese fisuresJorGDP (aja^nat PPPs). These -ist^pdardi^d Jtyd£iSg3^lFExpress.a^etage--ban. projgortion of GDPjper head in each As a result IrefancTand New Zealand their position considerably, while Germany, Luxembourg, Norwa Switzerland all fall down the scale. The last column in Table 6 shows t of relative benefit levels, calculated paring disposable incomes of persons social assistance with the disposable of the same household type where th earning^ average maTeea7mngsT*"T proba biy unaěTslálčéTnTřHTTépTacm of individuals in each country who ally receiving social assistance, since be expected that such individuals v earning less - perhaps substantially It average male earnings if they gaint Average replacement rates for all h types reveal three groups of countffe top come__the Nordic countries Norway, the Alpine countnés~OFSw and Austria, ana~tRe~f;Jgtjie^Tands. A torn come all four states of the US/ "southern European countries (excep according to our present informatio tween are found the English-speaki tries (except the USA), Japan and the EU countries. Identifying social assistance re Country groupings Can any patterns be identified in thi.< difference? To a certain extent the> Journal of European Social Polit 34 IAN GOUGH ET AL. they are not simple and they depend on which aspects of assistance programmes are regarded aTThlTmosn^lienfr Ljid^meJ_j(JL292)^nd Ledemel and Schulte (1992) distinguish two criteria - whether or not social_wg!rlgireatment measures are attached to the receipt of assistance, and the degree of programme centralization - to arrive it a fourfold "classification of 'poverty regimes'.*-They find all four cases In Wěštěřň~Europe,the object of their study: (1) institutionalized - centralized, unified, separate fromsocial work - in the UK (2) ďJfferentiated^äii€gôhca\ schemes alongside a general scheme, locaUy administered but codified and distinct from sočiahwork - in Continental countries (Germany, Austria, Netherlands and Belgium) (3) residual - local, marginal and with a strong emphasis on socíál~control and treatment.-in the Nordic countries (4) incomplete differentiated - categorical schemes for the non-ablebodied with no or, in the case of France, very limited general assistance tied to social control and treatment - in the Latin countries including France. These patterns map quite well onto Esping-Andersen's three regimes of welfare capitalism, as reinterpreted by Leibfried (1993) to demarcate a fourth, 'Latin' welfare regime. Building on their pioneering approach, we introduce several other features of assistance ~\pTo^áTnmésT^and^lsobring[North America, _ Japan and Australasia into the analysis. Adopting the systems analysis of social programmes we distinguish between inputs, programme structures, outputs and outcomes. Of these we regard the following as the most significant measures for grouping countries: • extent and salience: expenditure on social assistance and numbers of beneficiaries • programme structure: centralization v. local variation, particularly re benefit norms rights to benefit v. discretion individual entitlements v. wider family obligations liberal v. tough means-testing and work-testing © outcomes: benefit levels and replacement rates. We begin by combining the first and last of these criteria: extent and OHt£owe&JEigure 2 presents one measure of each: the numbers of recipients as a proportion of national population as a measure of the extent or salience of "social assistance, and the index of average net__ disposable "incomes on socjS^šjg^ce^after housing co"š'tš~äs"ä"proportion of per capita GDP - äs a measure of generosity. A notable finding is that these two dimen- Recipients as % population >10% 3-10% <3% Standardised benefit >120% levels after housing costs Ireland, Australia Finland, Iceland, Netherlands, Denmark, Sweden 80-120% UK, New Zealand, Norway, Luxembourg, Canada France, Austria, Belgium <80% USA Germany, (Italy?) Switzerland, Japan Spain, Portugal, Greece, (Turkey) Figure 2 Typology of extent and generosity of social assistance Journal of European Social Policy 1997 7 (1) SOCIAL ASSISTANCE IN OECD COUNTRIES sions are not correlated across OECD countries, lne extensive social assistance pro-"gřamrňeš-imhe~English-speaking countries" deliver relatively generous benefits in Australia and possibly Ireland, and vety low benefits in the USA. There are similar variations in those countries with a modest reliance on social assistance. We can distinguish the Nordic countries and the Netherlands, with generous replacement rates, from north-central European EU member states with around average benefit levels. Among those countries where the salience of assistance is minimal there is a similar variation: Switzerland, with one of the least extensive social assistance programmes in the OECD area, records the highest absolute benefit levels (and is still above-average when standardized for GDP), yet similar programmes in southern Europe deliver very low benefits. \ It is difficult to identify a single indicator of programme structure, or even a small number, because the number of variables is so great, and in most countries there are different schemes with different characteristics. Instead we construct an exclusion index (see Table 5, last column and notes). This ranks countries according to their score along the five dimensions shown in the table(tpegree of centralization and uniformity of benefits, narrowness ofjres,ource_unit, generosity of income Ind assets-tests, and laclc oFčliscretiorp The lowerji country's score, the closer it matches the ideal of a citizenship right to social assist: Higher the score, the closer is the cot variablě^houseríold-based^nd disc system of means-tested benefits. The clusionary' assistance programmes a English-speaking world; the most 'e ary' in the Alpine countries, Turkey, and Italy. Figure 3 plots countries according mension and the size of their assist: grammes. It shows that a clear rel exists between large-scale program rights-based, inclusionary structures, likely explanations are that rising ni beneficianelTTrnpose more bureaucr ttnized methods of "operation, wTi ex£e^3SB^ÍÍdj^^ressuxe^ior„r tralized financing and audit. The exte tinization and inclusion is not a fu modernization (witness affluent Swi but of high programme demands for of reasons.12 Social assistance regimes Despite the manifold cross-country ' identified, we conclude that certai patterns can be detected and we sug distinct social assistance regimes, based on the three dimensions dist above:/extent/ programme structure ' Beneficiaries % pop: Exclusion score >10% 3-10% <3% Most inclusive: <25 New Zealand, Australia, UK, Ireland, Canada 25-34 USA Netherlands, Denmark, France, Luxembourg, Finland, Germany, Sweden 35-44 Belgium, Iceland, Japan, Pc Norway, Italy Greece, S Most exclusive: >45 Austria Turkey, .l Figure 3 Extent and structure of social assistance programmes Journal of European Social Poln 36 IAN GOUGH ET AL. erosity/ They represent our collective judgement and can claim no firmer basis than that at presentr Future work might explore cluster analysis, qualitative comparative analysis and other techniques to discern country groupings. Selective welfare systems (Australia and New Zealand) These countries are unique in that all benefits_are means-tested. There are several categorical programmes, nationally organized, inclusive and rights-based. The means-testing is carefully constructed and monitored and is implemented in a consistent way. Assets and earnings disregards, and final benefit levels, are relatively generous. The public assistance state (USA) The USA exhibits an ěxteňšTvěšeť of means-tested benefits arrangedTn Hierarchy of acceptability and sfigŕfíäV"~Assets-tests -are generally tough but there are inbuilt earnings disregards and work incentives in every scheme. Benefits are variable, and tend to be low both in comparison with other countries and in relation to domestic poverty lines. Procedural rights, however," are well entrenched. This picture must be qualified in certain states, including New York, most of New England, Minnesota, Wisconsin, California, Washington, Hawaii and Alaska. In these states benefit levels are around the OECD average, though other features of US assistance remain the same. those in Britain. However, it is moving towards a more integrated system. Canada is a -----federal state with^considerable regional vari- ations, but in other respects the Canadian Assistance Plan has had much in common with social assistance in Britain. Under the new funding arrangements coming into effect from 1996, however, the picture will change. i :' Dual social assistance (Germany, France, Belgium and Luxembourg) These countries provide categorical assistance schemes for specific groups, supplemented with a general safety net. Some local discretion remains but is constrained within a national regulatory framework. Assets-tests are moderately flexible as are earnings disregards but benefits are below average. German Sozialhilfe has, despite its federal-Land structure, moved closest to the previous, British model: it is geographically equitable, codified, rights-based and extensive. Rudimentary., assistance (Southern Europe and Turkey) National categorical assistance schemes cover certain groups, mainly elderly and disabled people. Otherwise there is local, discretionary relief provided by municipalities or religious charitable bodies. Means-testing is not especially stringent and, apart from Turkey, obligations do not extend beyond the nuclear family. Cash assistance tends to be integrated with social work and other services. Welfare states with integrated safety nets , (Britain, Ireland and Canada) Income Sup-porfmthe UK is a large, national, general programme providing an extensive safety net at around social insurance levels. When Housing ,,. Benefit is inčluďéd, BěňefTtTevels are above the 6 OECD average. Rights to benefit are relatively well entrenched äňcTthe means-test contains important disregards, with some work iricen-tives for people witfTchildren thröügTTlFamily Credit. Ireland is at first sight a mix between this and the antipodean pattern. There are numerous categorical assistance schemes covering a high proportion of the population with means-tests and entitlements on a par with Ĺ, Citizenship-bq.sed but residitaLßssistan^e (the Nordic countries - excluding Norway - and the Netherlands) These countries have a single general scheme with relatively high ben-efrríevěls. Though there are national regulatory frameworks (to varying degrees), the role of local authorities is substantial .and„J.inks_ "' with social wo^k_anjd_socjaL.caxe-pexsisU--Strict means-tests combine with a more individual víew~oř entitlements and citizenship_-:basfiil-ap=---j peals systems. A tradition of full employment and/or universal welfare provision relegated social assistance to the margins of social programmes until the late 1980s. Journal of European Social Policy 1997 7(1) m SOCIAL ASSISTANCE IN OECD COUNTRIES I Benefits are very low and, for some groups and geographical areas, non-existent. -7 Decentralized, discretionary relief (Norway, 'Austria and~Switzerlind) "Norway and the ATpffiě"rauňtries contain elements of both the Nordic and southern European models. Assistance consists of localized, discretionary relief, linked to social work and with wider kin obligations. Benefit levels are above aver-age yet relatively few people claim assistance. This is partry^dilě" to the recorcTof lull male ' /, , employment and partly to low take-up, which is attributable to stigma and the substantial powers of intervention accorded to social welfare workers. 9 Centralized, discretionary assistatice, (Japan) Japan is difficult to place in the foregoing groups, containing elements of the British and Alpine models. It has a long-standing, nationally regulated assistance system with onlyHmočteŕäre'Ťocär variations related to differences in living costs. Yet it shares with' the Alpine countries a wider'cöncept of family obligation and there is circumstaht!är~évidénce "ThaFstigma is pervasive. It is evident that the social assistance regimes suggested here do not map perfectly onto ^sping-Andersen^s_ typology of J^eflSre^ regimes, injartjcular onto tiis measures qfwel-fare statestratification (1990: ch. 3). His first, t ^-"-'tíBerar*^ group of countries comprises the EngtíšTřspeaking countries plus Japanand Switzerland. THEfcannot provide a framework for understanding social assistance programmes for the following reasons. First, the role of assistance is qualitatively different in Japan and Switzerland, so much so that they < Other factors appear to median tween Esping-Andersen's princi] state stratification and the soi regimes we have identified. Diverse systems, common destination? This last section briefly explore; debates and likely future set social assistance in OECD natio The English-speaking count Journal of European Soda 38 IAN GOUGH ET AL. range of issues in common, including the cost of assistance, work disincentives,-fraud and targeting. A variety of policy initiatives have been developed to tackle each of these. In addition, concerns with the extent of poverty continue to feature in political debate in away that is unusual in most other countries. In all countries, except the USA, the high level of unemployment is also a cause for concern. These patterns are to be expected given the high cost of programmes and the large welfare clienteles in this group of countries. However, the issue of the behayiquraIincejitiv-es_of_welr__ fare"-~tcTdiscourage marriage or remarriage" and toencourage teenage pregnancy and wel-fanTdependency - appear to be a defining feature of the stigmatizing and divided public assistance system of the USA, and have now led to Federal withdrawal from guaranteeing any kind of welfare safety net. The idea that assistance is creating a new 'underčTásš' is relatively aEšénFlň "Australia and^ NewJZealand (although 'welfare dependency' is gaining currency as an issue for policy debate), and while present in debates in Britain and Canada it has not achieved ideological dominance. The_ questions. of programme costs, fraud ancl incentives figure in some other countries, ňotabtý those with relatively high spending levels and numbers of recipients - Germany and the Netherlands - or where unemployment is forcing more people onto assistance, as in Sweden and Finland today. In France and Belgium, where unemployment is higher and benefits are lower, there is also an ongoing debate about poverty and the role of targeted benefits, along with active labour market programmes in alleviating it. There is here some overlap with the agenda of the English-speaking world. A variety of other concerns feature in contemporary debates over social assistance. In the northern European countries, outside Scandinavia, these include the rights of immigrants and asylum-seekers to assistance benefits. In Germany and Austria, the transfer of the costs of institutional and domiciliary care for the frail elderly from social assistance to social insurance has also been a live issue. In all-Scandinavian countries a major question concerns the future role of social workers in the delivery of cash benefits: growing numbers of claimants have posed strains on the individual case-worker model and initiated a debate reminiscent of that in Britain 50 years ago. There is also a growing consensus in all these countries, except Norway, that more national regulation and uniformity should override traditional local autonomy in social assistance. The same is true, to a lesser extent, in Switzerland. Lastly, in Italy, Portugal and Spain, there is still another agenda, concerning the" practicalities of introducing a national safety netjor the first time - in part a response to the(J.992 European Council Recommendation on Sufficient Resources (92/441/EEC)ľ) However there is little sign that this is a salient issue in Greece. Though there are disagreements, most analysts agree on at least four changes affecting the environment of modeim "welfare states TöECDr T98 8f Pf aller et alľ, 199 í; Pierson, 1991; Taylor-Gooby, 1991; Esping-Andersen, 1996, ch. 1). 1. Economic. The first concerns changes in the economic environment, including the glob-alizatioďóf production, the rise of east Asian" economies, de-industrialization, exposure to global financial markets and (for some) the end of states' capacities to design their own political economic regime. A commonly perceived casualty of these shifts is traditional full employment and the emergence of atypical work. Moreover, pressures have been mounting to cut labour,..cgsts, both direct and indirect, in advanced economies, with impacts on taxation and public spending. 2. Demographic. The ageing of the population and upward pressures on welfare costs are familiar themes in international overviews. As noted earlier, older people do not make up a substantial proportion of assistance recipients in most countries because of alternative forms of provision. Where they do, it is often women who predominate, both because of Journal of European Social Policy 1997 7 (1) SOCIAL ASSISTANCE IN OECD COUNTRIES their longer life expectancy and because they are less likely to have adequate work-related insurance coverage. 3. Social. Fundamental shifts have occurred in the role; of: women which impact onjboth the labour market arid family structure. The ríšeTôf divorce, roňejparent families and other non-traditional family forms are examples of common trends, despite significant structural differences across nations.^ 4. Political. One set of political changes concé7ňT"'ffie^ffěčT~of~ťtá^ cJ2n^HO^|©^SÍ£^Íy^fÍ.ôrA,Ŕä.tterns: it is" contended that these undermine political support for traditional universal programmes in favour of differentiated services. Another school oJFtneóřištš"emphasizes the effect of growing inequality and new divisions__in» society on olderjpatterns ofjoHdarity, which create an impetus for the more affluent to exit from state programmes. Either way, it is argued, political support for inclusive patterns of social provision ebbs. Our analysis suggests that the first set of factors has had the most direct impact on the various social assistance regimes identified above. First, rising unemployment, and particularly long-term unemployment, drives more people t^IaTm "social assistance" as their entitlements to unemployment insurance benefit (where they exist) are exhausted. However, we find significant country differences in this process. Not surprisingly, we do not find such pressures in those countries with better job records, including the USA, Luxembourg, Norway, Switzerland, Austria and Japan (and Finland and Sweden before 1990). Even in countries with rising numbers of long-term unemployed, the impact on claimant numbers differs. Worst hit have been Canada, New Zealand, Britain, Ireland, France, Germany, the Netherlands and Belgium during the 1980s, and Sweden and Finland in the 1990s. But this is not the case in Denmark, where insurance-based benefits have been relatively effective in providing for longer-term unemployment, or in the Mediterranean countries where the long-term unemployed is ve 'end of full employment', if sue differently according to both and welfare regime. Second, concern has grown programmes create disincentive anoTcreate an 'unemploynient'r'r beneficiaries. This injurn exac gloyment,.and, boosts, state- sc Such concerns have stimulated ways to make the welfare sysť patible with changing labour i proposals can be loosely group< familiar^ categpxies__. 'ca rro ts' 'Carrots' include reducing the v of benefits as earnings rise, | cation, training and work e: grammes for jobless claimants, child care and other benel claimants with caring responsil bine these with paid work. 'Stic hanced surveillance of able^boi stricter.JKQrk^tests and sanction pliance, and reductions in r levels. It would appear from a bates in the Engjish^spsaking/ set of issues has dominated dis the future of social assistance. Again we find no universal tr bates overwork"dlslřicentives ; posals to rectify the situation major forms of assistance re above. They can be grouped in this purpose. ;1C Those countries with a pa or near-full employment: th tries, Switzerland, Austria a work incentives have not bet political debate (though the as such in Denmark Interestingly, in all of them c provided that their income < the permitted level. Such ] have occurred in the Norď primarily of the carrot type. Journal of European Soc 40 IAN GOUGH ET AL. care provision, the Youth Allowance programme in Denmark and the 'extraordinary workJ programmes in Norway-.------------------------ ( 2.] The limited social assistance regimes of southern Europe. Here the debate on labour-market disincentives is largely irrelevant. Social assistance for the able-bodied of working age is vestigial and its irrmact on local labour jnatkets ..slight. This is not to say tnat issues of work incentives are not relevant to certain unemployment insurance programmes, such as that of Spain. There has been experimentation with work and integration schemes in Italy, Portugal and Spain, but implementation has been patchy. 3. The remaining EU member states (excluding Britain and Ireland). Here a growing concern with the 'new poverty' and social exclusion in the 19$5s fuelled experiments with 'integration' programmes. The Revenu Minimum dTnsertion TRMI) in France, Minimex in Belgium, the reintegration programme in the Revenu Minimum Garanti (RMG) in Luxembourg all targeted young unemployed people and tied improved benefit levels to "insertion in training and work experience schemes. However, this combined carrot and stick approach has'Ti'äcľbnly a marginal impact, and the issue appears to be of relatively low political salience, except in France. 4'. The extensive social assistance states of the English-speaking world. It is in these countries that the relation between assistance and the labour már1cH'lYá's'"'áTšUmed greater importance in policy debates.13 New Zealand alone has opted solely for a variety of measures to make claiming less attractive, notably absolute cuts in benefit levels especially for unemployed claimants. The remaining countries have adopted a mixture pf carrot and stick. The carrotsHncTudeT lrigrier~benefiTs-T:oupled with activity-tests and partial individualization of income support for couples (Australia); a big extension of Earned Income Tax Credit, more generous tapers arid help with child care and medical care costs for lone parents and others who take up paid work (the USA); extensions and extra disregards within —Family, Credit and a package^ of back-to-wbrk provisions (Britain); and, in Canada, 'employment enhancing measures', a new child tax benefit and time-limited wage supplements. The sticks have included more intensive requirements for job search activity in all countries in this group. In addition, Britain has recently introduced the new Job_ Seekers Allowance, which further reduces trie insurance" element of unemployment__ protection and enhances work-seekirig^rer quiremehts. In the USA debate and policy proposals have become more punitive, leading to recent plans to 'end welfare as we know it'. Overall, the pattern of response to common economic pressures is complex. In many respects the differences are to be expected, given national differences in labour markets, employment records, social assistance regimes, political alignments and broader institutional traditions. The_ prominence of.work incentives in public debate is greater In countries with greäter~rélľä"nce on social assistance. Yet even within this group, differences are noticeable between the Australian selective welfare state and the American public assistance state. We find, therefore, a close correspondence between our classification of social assistance regimes and the policy responses to concerns about work and the labour market. Common global pressures generate differential national responses reflecting, among other factors, different social assistance regimes. Conclusion This article argues that means-tested social assistance schemes have in recent years acquired an increasing importance in the vast majority of industrialized countries - an importance which has not been reflected in the compara- Journal of European Social Policy 1997 7 (1) SOCIAL ASSISTANCE IN OECD COUNTRIES tive literature on welfare states. The article draws on a cross-national study of social assistance schemes in the OECD countries, which has attempted tö fill some of the gaps in comparative knowledge in this area. National assistance schemes are marked by substantial variation, which is clearly linked to countries' distinct and specific histories. Nevertheless there are also areas of similarity or convergence. We have selectedHT'niuTuöer of key criteria by which to attempt a grouping of different countries and on this basis identify eight broad regime types. Finally we consider the current pressures and debates which are driving policy change, concentrating on the debates on unemployment and labour-market change which seem to underlie many of the recent developments in social assistance policy across the OECD countries. What we have not done is offer an explanation for the patterns we observe. Elsewhere, we sketch an explanatory framework deploying social stŕučtíiŕaľ factors (modernization, family structures and employment structures) ancT^6IiHčH-iristítutíonarfactors (legacies of pTw^lľe^ffle^ernfrge'hcé' ofcafěgbncarásšiišt-ance and the postwar development of mainstream income maintenance) (Gough, 1996; Gough and Eardley, 1996, part 4). Nor does this article tell us how effective particular assistance configurations are at relieving poverty, achieving social integration, supporting work incentives or any other possible policy objectives. The research does, however, bring together a range of new data which can support further, necessary work of this kind. Notes Thanks to all the national informants and to the two referees and participants at the ISA Research Committee 19 annual meeting in Canberra, August 1996, for helpful comments on an earlier draft. 1 With apologies to Francophone Canadians, 'English-speaking countries' include Canada throughout this article. 2 The only work to our knowledge which provides a broad comparative analysis of sr that by Lodemel and Schulte (19' (1992). The contributors to Wal studied assistance schemes in soi providing useful information on I and outcomes. Room (1990) has EC member states. Both Rein (1 Jones (1985) offer insights on the of means-testing without system these to the analysis of specific coi Kamerman (1983) have estimate combinations of income support f assistance benefits, on the income families; a technique repeated by cently by Bradshaw et al. (1993) (1995). The contributors to Flora mass of systematic information o countries. Veit-Wilson (1993) eleven-country study of the settii minimum income standards. De S vides an original interpretation o social assistance in several Europi the USA. Fridberg and his collea compared social assistance in the Evans (1996a, 1996b) has made a means-testing in Britain, France a Oorschot and Schell (1991) have means-testing to the classic sociol social integration and exclusion. Other chapters and articles arisii have looked comparatively at par social assistance, including mean: 1996a) and work incentives (Ear at assistance in southern Europe > the Nordic countries (Bradshaw . coming) and from an Austr. (Whiteford and Eardley, 1997). 1 planned of the linkage between arrangements and the historica social security provision. The pn a general broad survey of the ma data collected for the study. 'Directly' restricts this to the pr< non-cash benefits by state agenci methods include tax allowances ; vate benefits or services. 'Individi means that services targeted on cations are excluded, though ger into the definition of the conti' benefit or service is designed to i Since a tied benefit, such as a h can be paid in cash, this term than referring to 'in-kind' or 'no Where different schemes are mut have added together the numbi Where they are not, we have tal recipients of the most extensive mum measure of the extent ol Journal of European Soci 42 IAN GOUGH ET AL. country. This table takes no account of the level of Feňefits^rčvvičledrwhetheř afamily isreceiving£50-------- or 50p a week, its members will be included. Similarly, the duration of assistance benefits will not feature: a benefit unit will be counted whether it receives a one-off payment or regular payments throughout the year. 7 Iceland falls into this group only because of the large numbers of older or disabled people receiving means-tested supplements to pensions. Some other countries have guaranteed minimum pensions which are in effect resource-tested but are not always included in the data supplied here, so there is some potential inconsistency. 8 To make the figures comparable, Sozialhilfe expenditure on residents of institutions in Germany and Austria is also included. 9 This method was pioneered by Kahn and Kamerman (1983) and has been further applied by Bradshaw et al. (1993, 1996) and Ditch et al. (1995). 10 Further discussion of the advantages and limitations of this technique can be found in Eardley (1996c). 11 Single persons aged 35, couple aged 35, single person aged 68, couple aged 68, couple with one child aged 3, couple with one child aged 7, couple with two children aged 7 and 14 years, lone parent with one child aged 3, lone parentt with one child aged 7. 12 When exclusion-inclusion is plotted against benefit levels across countries, no relationship emerges. 13 Ireland is an exception, with few proposals or policy changes in this area. References Atkinson, A.B. (1989) 'Social Insurance and Income Maintenance', in A.B. Atkinson, Poverty and Social Security, pp. 104-24. Hemel Hempstead: Harvester Wheatsheaf. Bradshaw, J. and Terum, L-I. (forthcoming) 'Is Social Assistance Special in the Scandinavian Countries?', Scandinavian Journal of Social Welfare. Bradshaw, J., Ditch, J., Holmes, H. and Whiteford, P. (1993) Support for Children: A Comparison of Arrangements in Fifteen Countries, Social Security Research. London: HMSO. Bradshaw, J. et al. (1996) The Employment of Lone Parents: A Comparison of Policy in 20 Countries. York: Family Policy Studies Centre. Burgess, R. and Stern, N. (1991) 'Social Security in Developing Countries', in E. Ahmad (eds) Social Security in Developing Countries, pp. 41-80. Oxford: Clarendon Press. Castles, F. and Mitchell, D. (1991) 'Three Worlds of ----------Welfare Capitalism-orJour?!,_A_ustraiian National^ University Discussion Paper 21, Canberra. Cornia, G. and Sipos, S. (1991) Children and the Transition to the Market Economy: Safety Nets and Social Policies in Central and Eastern Europe. Aldershot: Avebury. de Swann, A. (1988) In Care of the State. Oxford: Polity Press. Ditch, J., Barnes, H., Bradshaw, J., Commaille, J. and Eardley, T. (1995) A Synthesis of National Family Policies 1994. European Observatory on National Family Policies: University of York. Eardley, T. (1996a) 'Means Testing for Social Assistance: UK Policy in an International Perspective', in D. Coyle, N. Lunt, P. McLaverty and J. Sorensen (eds) Welfare and Policy: Issues and Agendas, pp. 58-77. London: Taylor and Francis. Eardley, T. (1996b) 'From Safety Nets to Springboards? Social Assistance and Work Incentives in the OECD Countries', Social Policy Review No. 8, pp. 265-85. London: Social Policy Association. Eardley, T. (1996c) 'Lessons from a Study of Social Assistance Schemes in the OECD Countries', in L. Hantrais and S. Mangen (eds) Cross-national Research Methods in the Social Sciences, pp. 51-62. London: Pinter. Eardley, T., Bradshaw, J., Ditch, J., Gough, I. and Whiteford, P. (1996a) Social Assistance Schemes in OECD Countries: Volume I. Synthesis Report, Department of Social Security, Research Report 46. London: HMSO. Eardley, T., Bradshaw, J., Ditch, J., Gough, I. and Whiteford, P. (1996b) Social Assistance in OECD Countries: Volume 11. Country Reports, Department of Social Security, Research Report 47. London: HMSO. Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism. Cambridge: Polity Press. Esping-Andersen, G. (1996) Welfare States in Transition. London: Sage. Eurostat (1993) Digest of Statistics on Social Protection in Europe. Volume 4: Family. Brussels: EEC. Evans, M. (1996a) Means-testing the Unemployed in Britain, France and Germany, Welfare State Programme Discussion Paper WSP/117. London: LSE/STICERD. Evans, M. (1996b) Families on the Dole in Britain, France and Germany, Welfare State Programme Discussion Paper WSP/118. London: LSE/STICERD. Flora, P. (ed.) (1987) Growth to Limits: The Western European Welfare States since World War 11, Vols 1-4. Berlin: Walter de Gruyter. Fräser, N. (1987) 'Women, Welfare and the Politics of Need Interpretation', Thesis Eleven, No. 17. Fridberg, T. (ed.) (1993) On Social Assistance in the Nordic Capitals. Copenhagen: Nordic Council of Ministers. journal of European Social Policy 1997 7(1) SOCIAL ASSISTANCE IN OECD COUNTRIES Gough, I. (1994) 'Means-testing in the Western World', Richard Titmuss Memorial Lecture (May), Hebrew University of Jerusalem. Published in Hebrew in Social Security, no. 43, January 1995: 24-44. Gough, I. (1996) 'Social Assistance in Southern Europe', South European Society and Politics 1 (1): 1-23. Gough, I. and Eardley, T. (1996) 'A Comparative Study of Social Assistance in OECD Countries', paper presented to the annual meeting of Research Committee 19 of the International Sociological Association (August), Australian National University, Canberra, Australia. Jones, C. (1985) Patterns of Social Policy. London: Tavistock Publications. Kahn, A. and Kamerman, S. (1983) Income Transfers for Families with Children: An Eight Country Study. Philadelphia: Temple University Press. Leibfried, S. (1993) 'Towards a European Welfare State', in C. Jones (ed.) New Perspectives on the Welfare State in Europe, pp. 133-56. London: Routledge. Ledemel, I. (1992) 'European Poverty Regimes', paper presented at the International Research Conference on Poverty and Distribution, Oslo. Lodemel, I. and Schulte, B. (1992) 'Social Assistance: a Part of Social Security or the Poor Law in New Disguise?', paper presented at the conference Social Security Fifty Years after Beveridge (September), University of York. Organisation for Economic Co-operation and Development (1988) The Future of Social Protection. Paris: OECD. Pfaller, A., Gough, I. and Therborn, the Welfare State Compete? A Comf Five Advanced Capiialist Cow. Macmillan. Pierson, C. (1991) Beyond the Welf an Political Economy of the Welfare Si Polity Press. Rein, M. (1983) From Policy to P> Macmillan. Room, G. (1990) 'New Poverty' it Community. London: Macmillan. Sainsbury, D. (1993) 'Dual Welfare anc of Access to Social Benefits: Incoi Policies in the UK, the US, the T Sweden', Journal of Social Policy 22 Taylor-Gooby, P. (1991) Social Chang and Social Science. Hemel Hemp: Wheatsheaf. van Oorschot,W. and Schell, J. (1991)' Europe: A Growing Concern', in M. Sociology of Social Security. Edinh University Press. Veit-Wilson, J. (1993) 'Comparing Minimum Income Standards', pap conference of ISA Research (September), Oxford. Walker, R., Lawson, R. and Townsenc Responses to Poverty: Lessons from Heinemann. Whiteford, P. and Eardley, T. (1997) '' Net: Public Assistance in OECD C Policy Research Centre Discussion University of New South Wales. Journal of European Socio THREE Yves Bonny and Nicoletta Bosco Within the framework outlined in Chapter Two, this chapter focuses in particular on the income support measures for the able-bodied poor adopted in the countries and cities examined. Our aim is twofold: firstly, to reconstruct in detail the mode of construction and implementation of the measures in order to understand their impact and significance for the life chances of beneficiaries; and secondly, to develop a basis for comparison, allowing us to go beyond the specific features of each case study and integrate them within a more general picture. In order to do this, we need to take into account the multiple dimensions and levels involved in the actual implementation of a programme. These include the selection of the beneficiaries, the perspectives and options offered by the programme to those involved and its consequent social significance. The only way to understand the effective range of a measure and its impact on poverty is, in fact, to reconstruct the whole set of regulations applied in the social treatment of poverty. Three levels of contextualisation are crucial. The first, and most obvious one, has already been dealt with in Chapter Two. It concerns the socioeconomic and social-demographic background within which the measures are developed and implemented. This plays a decisive role, most notably in the employment prospects offered and the selection of beneficiaries. With respect to the dimensions mentioned above, it contributes to the definition of the profile and the image of those affected by income support measures. The profile will differ considerably according to the dominant forms of economic activity, whether the city is flourishing or in dechne, the structure of the population by age-group, the importance of immigration, the attractiveness of the »I Social assistance dynamics in Europe city for surrounding populations, and so on. At the same time, the economic conditions shape the perspectives and options open to the actors. The accompanying programmes included in the various systems of social protection - including training courses, funded job schemes, and so on - obviously cannot be isolated from the employment situation specific to the local context, that is the level of unemployment, types of job sldll in demand, wage levels, the possibility of woiidng part-time, and so on. The strategies developed by the beneficiaries of income support programmes, and the way that they experience their situation, will depend on the distribution of poverty and the employment prospects of the city concerned. Second, income support measures take place within a specific cultural and political universe, which plays a major role in the social construction of poverty and the response provided. Poverty has no univocal interpretation, inasmuch as we are not dealing with a reality independent of a social framework. As mentioned in the introduction, the very notion of poverty involves a number of social representations, both in relation to the qualification of situations and persons and the definition of the forms of social intervention considered appropriate in face of the acknowledged poverty problems. No objective standard of living indicator can tell us whether or not a given person will be regarded as poor by his/her community, if support will be provided and if so how, if this will lead to Stigmatisation, and so on. In order to know this, one needs to reconstruct the whole web of meanings that shape the notion of poverty in a given social universe. This will depend on prevailing economic conditions, but also on political structures and dominant cultural and ideological representations. Thus, if we speak of exclusion in some countries, the underclass in others and marginalidad in others still (see Fassin, 1996), this is not simply a question of terminology, but refers to three very different social constructions of poverty. One cannot study poverty and resulting measures independently of the social universe that names it and gives it meaning. This is always social, historical and dependent on the systems of reference1 that guide those involved in the struggle against poverty, from the politicians who pass the laws to the social workers who intervene in concrete situations. In this regard, there are very significant differences among the countries and cities studied. Thirdy-one cannot understand-die-exact range of a programme unless one locates it within the whole social protection context where it takes place. By this we mean not simply the public system of social protection, but also informal forms of protection connected with primary solidarity. These two modes of protection are not independent. We therefore need Income support measures for the poor in European cities to understand how primary and secondary solidarity2relate to each other, at the cultural and ideological level, the institutional and legal level and also the practical level, since these will shape the distribution of responsibilities between public actors, associations and interpersonal and family solidarity. The different configurations translate into cognitive and normative models that structure the actions and representations of the actors and define their respective expectations. "We need in particular to understand the exact scope of the measures and the way they fit into other programmes. Each system of social protection forms a whole, so most of the characteristics of a single programme cannot be understood in isolation. As mentioned in Chapter One, a specific measure finds its public and its effective range partly through the way it relates to other elements of the public social protection system (for instance, unemployment allowance or family allowance), and partly through different regulations (for instance, regarding the minimum wage). Thus, in different policy contexts, the able-bodied 'public' benefiting from social assistance income support will be smaller or larger, with greater or lesser disadvantages in relation to the labour market, and even a more or less accentuated specific gender and/or age profile. It will depend not only on the national and local social and economic situation, but also the overall system of income protection. This in turn has important consequences for the impact of a given measure. There is, in our view, a considerable advantage in focusing on the local level, without necessarily embracing a 'localisť perspective. In particular, it avoids imposing an inappropriate concept of'welfare state'. This is especially important in countries in which social policy is not standardised at the national level, and where there are consequently still large regional or even municipal variations. It allows us to examine measures belonging to very different institutional forms side by side. It also avoids stressing the contrasts that may exist between national and local regulation of poverty on the sole basis of the formal organisation of the schemes. In fact, aside from the official institutional set up, even in countries where social policy is not highly centralised or standardised, there may nevertheless be similarities between cities for cultural and historical reasons, or due to the economic and political situation. These factors can play a major role in the construction of homogeneity at the national, or at least the regional-level3. - ____________—______________— - ____________ Conversely, in countries endowed with uniform national institutional schemes, the 'local' approach allows us to emphasise the existence of variable modes of implementation and very different degrees of mobilisation of actors. The local level is not simply a locus for the Social assistance dynamics in Europe application of rules and bureaucratic procedures whose meaning is obvious. It plays a crucial role in the way a measure, even though centralised, is concretely translated into the actions, decisions, relations, resources, and so on, which define -a-certaintype- of -intervention.- - Finally, the local level is particularly relevant in the present context of devolution, especially in connection with integration measures, which are increasingly being used in income support programmes4. The ambiguity of this evolution has been underlined by a number of analysts (for example, the European Foundation for the Improvement or Living and Working Conditions, 1999). The decentralisation of public action, motivated by the greater flexibility achieved, the personalisation of follow-up and mobilisation of local actors and resources, may actually engender major disparities. This is due to the great variations that exist in the investment by local actors, particularly locally elected representatives, in the implementation of integration policies5. Analysis at the local level is therefore essential to permit the examination of the network of actors concerned and the logic of their intervention. It enables us to bring to the fore any blockages or synergies that may explain those disparities, as well as the types of supra-local regulations which help to preserve uniformity in the programmes and equity of treatment despite decentralisation. The local welfare configurations approach implies that the systems have been analysed and compared on the basis of their actual patterns of implementation, rather than just their formal and institutional characteristics. We have tried in particular to identify any divergences between legislative or statutory texts and decisions taken in the field. This enables us to determine the actual treatment of applicants and beneficiaries in the cities studied, and thus evaluate not only the differences between the cities, but also the sources of these differences (official texts, professional cultures, local political orientations, and so on)6. Flomogeneity at the legal level can hicie profound differences in the way policies are implemented de facto and similar outcomes may arise from very different processes. To take a particularly revealing example, while the leaving curves for Milan and Helsingborg are surprisingly similar, the reasons are very different. In the former, the rapid exit from the programme can be attributed to the scanty benefits and their limited duration, while in the second it is due to a particularly generous benefit scheme coupled with numerous programmes for employment assistance and the development of professional qualifications. In other words, the meaning of'entering' and leaving' a scheme is not at all the same in the two cases. We may conclude that these curves as such give us no indication about the nature 84 Income support measures for the poor in European cities of the measure or its effectiveness in making people independent and enabling them to get beyond the initial condition of poverty. In order to understand the way the policies really function, we also closely-examined the characteristics of the various measures ^ the institutional framework of which is synthesised in the Appendix. They were examined, as we have said above, both with regard to their formal, legal features, and at the level of their implementation. Further, the beneficiaries' perspective was also taken into account in order to assess the actual impact of a given measure on their lives. Through a number of interviews conducted in each city with former and current beneficiaries7, we tried to capture their objective life conditions within each local context, as well as their subjective perception of their poverty situation, the measure they were (or had been) benefiting from and the social services that dealt with them. In order to analyse such diverse material and fully exploit the potential of our approach, we developed a multidimensional model constructed on an inductive basis from the 13 cities studied". Unlike analytical models that focus on purely formal differences, or those deduced a priori without practical verification, our objective was to identify differences from within. We therefore did not try to develop a typology as such. Of course, the eight dimensions selected in our model are not independent of each other, and it is possible to identify certain polarities representing typical forms. For instance, familiar contrasts, like that between Northern and Southern Europe, can be recognised. It is also possible to rank the cities, for instance, according to the degree of institutionalisation of policies against poverty. This being said, it seemed important to preserve as much as possible the full diversity these dimensions contain. They can too easily be obliterated in a typology, which tends to rigidify contrasts, and group together in an often questionable way heterogeneous dimensions in a single type, forcing all the cases into a limited number of configurations. On the basis of a detailed analysis of the local contexts, presented in Chapter Two, we have tried to stress the diversity of possible positions on each dimension studied, whatever the convergence that can be established in other respects. The preference given to a comparative structural analysis, as opposed to the construction of a typology, in the strict sense of the term' also allows us to bring out original features of each local welfare configuration. For example, if one were to draw up a typology on the basis of the degree of institutionalisation of policies against poverty, Barcelona would lie in an intermediate group, together with other cities that are 'in transition'. However, this would fail to catch one of the original characteristics of 85 Social assistance dynamics in Europe the Catalan experience, which is the attempt to draw on the collective sense of solidarity to activate the proximity resources (that is, neighbourhood, formal and informal community associations) that are still very much alive in Spanish society. This is particularly significant in view of the many problems generated by the 'negative individualism' associated with highly centralised social protection systems, built on a rigid distinction between secondary solidarity, linked to public intervention, and primary solidarity (see Gauchet, 1991; Castel, 1995;Astier, 1997, on the notion of'negative individualism'). One of the numerous problems we had to face in developing a comparative framework was that there was not always a single programme applying uniformly to all situations of poverty and treating them in a uniform way. In many cities, treatment varies according to the categorisation that is applied to an individual or a household. This can sometimes mean that different measures apply to different groups of people. It is difficult to ignore all these variations, as they are quite often significant. A closer analysis, however, showed that beyond the very real differences in treatment according to the category of claimants, these measures also have a number of cross-category characteristics. These were important enough to allow the identification of a common underlying model of income support also in the cities that use a categorical approach. The category-based distinctions are certainly a crucial dimension of this model, but they do not spoil its coherence, which justifies a general comparative analysis (obviously this assumption can be misleading, especially in contexts that have high scores concerning discretionary power). In the following paragraph, we present the different dimensions of our structural analysis. After this, we describe the welfare measures in the 13 cities in the ESOPO project, trying to combine an emphasis on the specific local features with a comparative analysis. Chapter Three closes with a series of diagrams representing the different local configurations on the basis of the selected dimensions. These complement Figure 2.2 in Chapter Two, in so far as they include more dimensions, but focus only on income support measures9. Tlie various dimensions of income support For the sake of simplicity, we have used a model in which a limited number of dimensions are considered, although we try to take into account all relevant sources of variation. In a number of cases, this has forced us to put together within a single dimension aspects that were analytically distinct. When this is the case, we have indicated the different aspects Income support measures for the poor in European cities involved and specified in the next section (Local patterns of income support compared) how these affect each city. The major dimensions in our study concern three aspects: (a) the dominant orientation within which the measures can be placed, (b) the conditions of access to and protraction of economic support, and (c) their content. The local welfare configuration of each city is represented graphically in Figures 3.1-3.6 at the end of the chapter. Scores have been attributed to each of the selected dimensions, and a local welfare 'shape' drawn in the form of a polygon, constructed by finking the scores for each dimension. Since the dimensions selected are themselves multidimensional, the scores have to be considered as combined qualitative evaluations of each dimension. Dominant orientations Bureaucratic regulation versus discretionary power Bureaucratic regulation concerns decisions taken in strict pursuance of the rules/laws (whether local or national), whereas discretionary power refers to the margins of manoeuvre allowed to social workers in granting access to benefits. This may include the choice of the kinds of measure applied, the duration and amount allowed, and the renewal and protraction of benefits. Discretionary power may result from the lack of clear rules or be actually written into the regulations in acknowledgement of the complexity and diversity of need. Discretionary power can thus take on very different meanings within the local contexts. Social workers can sometimes have a wide margin of manoeuvre in assessing the needs of people claiming economic benefits, apart from the categorisation of need provided by the law. But discretionary power can also be due to the absence of a unifying legislative and/or regulative framework. This is sometimes replaced by sub-local non-written rules employed by social services, resulting in a high degree of variability of decisions, even within a single social work office. Discretionary power and bureaucratic regulation characterise general orientations typical of two different cultural models. On this basis, we can position each local welfare configuration on a continuum (see the horizontal axis in the diagram of Figure 2.3 in Chapter Two) ranging from a highly bureaucratic approach, where interaction and negotiation play a limited role (as in Helsingborg and Gothenburg"1) to an approach in which negotiation, and even personal relations between social workers and perspective beneficiaries are crucial for granting and obtaining support (Lisbon and Cosenza). In Cosenza this sometimes leads to what can be Social assistance dynamics in Europe defined as a patronage system. This polarity can also affect variousaspects of benefit allocation, including access and the renewal and duration of benefits, but also their contents and generosity, in so tar as social workers are allowed to choose among measures that can differ in length or generosity. The score attributed to each city in our sample reflects the answers to the following questions: does the claimant, having the required characteristics for each local context, have guaranteed access to income support, or can social workers apply their personal opinion regarding the deservedness of the individual? Can they use their discretionary power for controlling the access (or renewal)? Can they choose among different measures or are they legally bound to apply given rules? Cities where the prevailing model of intervention is one of bureaucratic regulation will have higher scores in this dimension. Lower scores mean a high degree of discretionary power in applying the rules. Family obligations versus collective solidarity The poles of family versus collective solidarity (see the vertical axis in the diagram in Figure 2.3 in Chapter Two) represent two very different models of solidarity. Family orientation implies a model where family obligations and responsibilities make up for the lack of adequate public services, or where public intervention is admitted only when the family and its safety net are not sufficient, or defined as being inadequate. Collective solidarity, on the other hand, refers to situations where rights are highly individualised and guaranteed by universalistic procedures and public obligations. This dimension also includes the regulation of so-called 'obliged kin' (for example, the extended family members who have legal responsibility for providing financial or other support) and, more generally, the relevance of family (including kin) obligations within each context. Where family-kin members are still considered the main providers, public support represents the last resource to which individuals and families can turn in case of need. It is well known that one of the strongest features of welfare systems in Southern Europe is the central role assigned to the family in supporting individuals well beyond childhood, and sometimes irrespective of cohabitation. Once again, the orientation with regard to solidarity has implications for both objective and subjective experience. While giving primacy to collective public solidarity has the advantage of ensuring everyone minimal protection, it also tends to discharge the family group from any sense of responsibility with regard to its members. This in turn can generate a 88 Income support measures for the poor in European cities kind of'negative individualism', in which a person's poverty is combined with a lack of social ties. Conversely, although giving primacy to family solidarity may facilitate social integration, it leaves part of the population unprotected and may also result in an overall impoverishment of the family when it is crushed by multiple demands for support that it has not sufficient resources to meet. This can become even more problematic when family relations are strained or non-existent. In this connection, we can observe real discrepancies between laws and behaviour, particularly in northern Italy. Here the legal priority given to family solidarity is increasingly in contradiction with cultural transformations tending to individuaUse relations within the family. Thus, we have assigned a low score to Turin because, even though the responsibilities of'obliged kin' are defined by a national law, and further enforced by a local law, they are Hot usually applied by social workers when they involve non-resident kin. The medium value assigned to Halle means that this condition is applied only sometimes, and at the social worker's discretion. Universal versus category-based orientation Regarding the various anti-poverty measures, there is an important contrast between those with a universalistic orientation and those that target defined groups of able-bodied poor on the basis of a notion of deservedness or priority. Targeting may coexist with a high degree of national and local institutionahsation, or vice versa a high degree of fragmentation and discretion. Thus, this contrast may not be adequately captured in the bi-dimensional diagram Figure 2.3 in Chapter Two. Universalism does not imply flat homogeneity, and within the universalistic orientation we do not necessarily find uniform treatment. Distinctions may be made between different categories of applicants or beneficiaries; but the principle behind such distinctions is 'to treat different cases differently', without any evaluation of deservedness being implied. By contrast, in the category-based orientation there are explicit or implicit evaluations of deservedness/undeservedness. Of course, when we consider concrete cases, the distinction is rarely as clear-cut as that. Even in places where a minimum level of support is considered a right, we often find an implicit evaluation of deservedness/undeservedness, especially with regard to able-bodied poor. In France, for instance, non-normative evaluation is recent and corresponds to a retroactive interpretation of the category-based logic applied before the RMI (Reuenu Minimum d'lnsertion, or minimum income for integration) was introduced. The philosophy followed then was aimed Social assistance dynamics in Europe at distinguishing the 'deserving' from the 'undeserving' poor. One had a right to social assistance if one was unable to work, such as an invalid, a disabled or older person, or was exempted from the obligation to work (as in the case of single parents with a child below the age of three; see Castel and Laé, 1992, pp 10-11). The introduction of RMI represented an acknowledgement that even able-bodied adults with no serious mental or physical disability preventing them from working, or without heavy caring obligations, could deserve support when in need. Yet, even today, there remains in France some notion of deservedness (for example, lone mothers of very young children, who benefit from a special measure) and undeservedness (young adults below the age of 25, who are excluded from RMI if they do not have children of their own). Conversely, in a case such as Turin, the fact that it is allowed to provide support 'according to need', even to individuals in officially less deserving categories", somewhat relaxes the categorical nature of the local system in the direction of a greater universalism. However, too strong an element of discretion in interpreting regulations may undermine this more universalistic approach, since there is too much scope for variability. Thus, the social assistance system in this city encompasses various, rather than a sole model of support (Bosco et al, 1999). People's experience of the situation differs profoundly depending on whether income support is explicitly associated with a concept of deservingness, or whether it aims to provide an appropriate form of support without imposing any value judgement. Whereas the former case is based on some notion of merit, the latter is based on the idea of a legal right12. In relation to the universalistic dimension, we assigned a higher score to cities where, within the programmes provided for the able-bodied, social workers were not allowed to choose between treatments of greater or lesser generosity on the basis of a notion of deservedness or priority. Conditions of access and protraction of economic support measures Selectivity Selectivity can refer to two different characteristics. The first concerns the existence of income thresholds and therefore implies assessment. In almost all the ESOPO cities people have the opportunity to claim for benefits if their personal (or household) resources fall below a certain level13. Within this means-tested logic, however, the resources taken into consideration for purposes of assessing entitlement can vary greatly'4. Income thresholds below which entitlement is granted may vary according to the target population, or the household composition (particularly its income support measures for the poor in European cities age structure). Or, in some places, they may be the same for everyone claiming benefits. Selection criteria have a profound effect on the characteristics of those who have access to the system. The more severe the means-testing, and the lower the income threshold, the more serious the problems of recipients. The following two chapters analyse in-depth how this affects both the characteristics of beneficiaries and their welfare 'careers'. The second aspect of selectivity results from the presence or not of a tight budget constraint. Inadequate resources can mean that the system is unable to protect people in need. The degree of selectivity will depend on how much funding has been allocated to the welfare programme, and will directly condition the rate of acceptance of applications. As a scarcity of resources leads to high selectivity, this limits the number of claimants who can accede to benefits, and risks undermining the principle of financial aid as a right. For example in Milan, even though municipal regulation established a principle acknowledging universal entitlement to a Minimum Living Allowance, the number of beneficiaries and amount of the benefit depend on the resources made available each year in the municipality budget. Consequently, the certainty of entitlement is subordinated to the discretion of social workers and the will of the governing political coalition. Moreover, the impact of selectivity on people's objective and subjective experience should not be underestimated. The more selective a programme — at both levels — the more negative the general portrait of the beneficiary population, and the more difficult it is for them to become independent and socially integrated. This explains why the duration of income support — when not set by administrative or legal rules, as, for example, in Milan — seems to increase with the rigour of selection. It is also why the beneficiary population's perception of the measure is likely to be more negative when selection is more rigorous, since the measure then seems like a form of charity for the most marginalised, inducing in beneficiaries a feeling of personal failure. Recip ien ts' du ties This dimension includes a range of aspects concerning the duties required of recipients of economic support and the way that these are handled by social workers. The kind of duties may differ considerably. They may involve simply the formal willingness to look for a job (that is, being enrolled at the employment agency), or the obligation to perform some kind of job — or to be trained for a job — in exchange for the benefit, or involvement in social integration/activation activities. As many authors Social assistance dynamics in Europe point out (European Foundation for the Improvement of Living and Working Conditions, 1999; Gough et al, 1997) there has been a shift in recent years towards the principle of workfare, even though there is no consensus regarding its definition15. There iš ä subtle distinction between 'pushing' factors (which suggest somehow punitive or disciplinary elements) and 'pulling' factors in the form of incentives or activation strategies (Heikkila, 1999). The logic on which such duties are based is twofold and varies according to the local context. On the one hand is the idea that receiving subsidies can produce passivity and dependency in the able-bodied poor and that these perverse effects must be hindered by introducing compulsory work training or job experience. On the other is the assumption that the able-bodied do not totally deserve support or, at least, that they must 'earn' it by behaving according to social expectations. Where this second attitude prevails, recipients become more indictable if they are unwilling to comply. It can sometimes happen (as inVitoria and Turin) that 'social integration' or 'activation' measures are profoundly ambiguous: for example, training programmes are offered more as a way of controlling the beneficiaries than for their relevance for professional integration. This engenders passive attendance, simply to avoid sanctions. There are differences not only in the demands made on applicants, but also in the way they are enforced, often depending on the category to which recipients belong. However, in general, the greater the degree of undeservedness perceived, the higher the pressure and control (that is, longer procedures for preventing fraudulence, more frequent home visits or contacts with social workers, and so on) exercised before granting benefits or prolonging recipiency. Some support measures entail no precise demands, either because the general profile of beneficiaries suggests that they will fail on the job market, or because there are few resources for setting up social and professional integration programmes. In these cases - Portugal is an example - such demands, when they exist, are left to the discretion of social workers, who can choose to use their networks within the informal economy to find reinsertion opportunities. Often recipients are 'obliged' to seek'reinsertion', in the terms defined by social workers and/or by institutional rules, as a counterpart for receiving benefits, with little room for negotiating what they are willing to do, or under what conditions. The main idea is that beneficiaries must 'do something in return' for social assistance. Benefits may be discontinued or reduced if this requirement is not fulfilled. The question is then how strictly the notion of obligation is interpreted and applied: what is an 'active'job search or a 'reasonable'job offer, what forms of pressure are on Income support measures for the poor in European cities exerted and what sanctions or penalties applied for non-compliance? Our research demonstrates that in the cities whose measures stress the notion of receiving'something in return', different positions are adopted - weak pressure in the two Swedish cities, steadier pressure in Halle. This has a strong effect on beneficiaries' objective and subjective experience. Since this dimension stresses the idea of'obligation', in attributing scores to each local context we have considered both the presence of an obligation (to work or to follow courses) and the extent to which benefits are cut (in amounts or duration) when recipients are unwilling to comply. Contents of the measures Generosity This dimension cuts across all the others, in so far as it involves the income thresholds, the equivalence scales, the amount of the benefit, its duration and the possibility of renewing it, as well as the presence of accompanying measures and linked benefits. The question of generosity is highly controversial within the European discourse on welfare reform. A general definition of adequate generosity could be that benefits should provide the possibility of staying above the poverty threshold. But this is far from being accepted everywhere, because of the idea that if benefits are too generous they could discourage people from looking for a job (see Geldof, 1999). Moreover, to appreciate the real impact of the sum granted, it is important to situate the measure in its social, economic and cultural context. This means taking into account the cost of living and also the state of the local job market - in particular, wage levels as compared with the benefit granted - as well as unemployment rates and the type of jobs potentially accessible to welfare beneficiaries. From this perspective, we might hypothesise that in Halle a relatively favourable benefit level, combined with not very enticing employment prospects, would encourage people to stay on welfare. But this does not automatically follow because, among other things, the cultural variable - in this case, the importance of paid work in constructing personal identity - plays a crucial role. The more one's self-image is built around paid work, the less one can envision oneself on permanent or long-term income assistance. In other words, the less one can settle into what might be called the 'welfare mentality'. Next, we need to evaluate what percentage of a beneficiary's total resources are represented by welfare payments. Here the relevant rules are those concerning the recipient's work-earned income. With the aim of facilitating access to full professional activity, certain measures make Social assistance dynamics in Europe considerable allowances for money earned through work, others do not. This of course has a considerable impact on available resources (and on the length of time benefits are paid). Then there is the possibility of obtaining undeclared complementary income in the 'informal5 economy. This varies from one local situation to another, depending on the local labour market, the size of the informal economy, and the degree of institutional control on the latter. Finally, there is the weight of private solidarity — assistance from the recipient's network of kin and friends or from charitable associations — compared to public solidarity. Once again, vre observe great variation among the different cities. In some, the welfare benefit constitutes the recipient's only resource (Sweden), whereas in others family solidarity and the informal economy play an essential role in most recipients' living conditions (Barcelona16 and Turin). All these considerations concerning the generosity of income support have been simplified in order to allow the definition of a single score for each city. Thus, in order to define this dimension, we have decided to use a very rough index that has been calculated on the basis of the answers to the following questions: e Will the receipt of income support enable a household (in this case with one member only) to reach the poverty line? - Does the equivalence scale used to equalise different household sizes stress individual needs or expect a large degree of household sharing? • Are housing costs covered? When the answers to these questions are affirmative, the score given is high. We have in fact assigned 10 points to the Swedish case. The scores of all the other countries (or cities) have been weighted in relation to the Swedish standard17. Duration Here we are concerned with the presence (or absence) of specific time limits on the receipt of ■benefits for individual recipients or categories of recipients (not the actual duration, which "will be discussed in Chapter Five). The variability of time spent on welfare is a very complex question, strongly influenced by the general features of local welfare systems and their cultural meaning, but also depending on labour market conditions, on patterns of family and kinship solidarity, and on the presence of charitable associations. Duration does not have a linear relationship with income support measures for the poor in European cities a lack of selectivity or generosity. In cities where beneficiaries are in situations of extreme poverty and the services themselves have only limited resources, low benefits sometimes go together with long-term duration (for example, Lisbon), even if the amounts are so small as to make no significant difference in the recipients' conditions. In other cities, such as Milan, low amounts can go together with short duration, and benefits may be interrupted after a defined period, even if the recipient is still in a situation of need. This does not occur in Sweden, France and Germany, and to some degree in Turin (see endnote 11). In these cities, in fact, once recipients have been acknowledged as such, they can receive benefits as long as their need persists. We have assigned a high score when there is the possibility of continuing to receive financial support as long as recipients meet the criteria governing access to assistance. Activation measures The concept of activation refers here to the positive non-financial incentives or opportunities offered to recipients, who are free to choose whether to participate or not in programmes that provide some sort of training, education, employment services or work (Heikkila, 1999). This dimension therefore involves the combined evaluation of a range of actions and programmes aimed at improving the situation of the beneficiaries, over and above income support. In this regard, we can distinguish between 'social aid' and 'professional integration' activities'". The evaluation of this dimension cannot be performed in absolute terms, and implies taking into account the kind of beneficiaries involved, as well as their perceptions. Given the orientation of the ESOPO study, wliich concerns the able-bodied, we decided to allocate greater weight to forms of action aimed at encouraging professional integration. While considering the average profile of beneficiaries and the local economic situation, we have compared the measures on the basis of the forms of activation in place for the able-bodied, and assessed to what degree they contribute to improving their situation and chances of overcoming poverty. A particular problem at this level is posed by the institutional and organisational set-up of the different welfare programmes, wliich vary considerably from one city to the next. The most significant distinction concerns the relationship between income support and training or employment programmes. In some cities the latter are integrated within the income support measure, in others they are separate — the social welfare office providing the financial support and the employment office Social assistance dynamics in Europe coordinating the programmes related to professional integration. For tliis reason, we have tried to evaluate the quality of activation measures on a global level, focusing attention not so much on the formal content of each measure, or the institutional and organisational set-up, as the whole set of programmes available to beneficiaries aimed at their professional integration'9. Three aspects of activation programmes seem particularly relevant: how much they ofier (beyond the monetary aspect), how effective they are, and how much personalised assistance they provide. The first refers both to the variety of help offered and its appropriateness for improving the recipients' situation, either in social and psychological terms, or in terms of training and professional qualifications. Effectiveness can be measured by looking at the usefulness of the programmes for getting beneficiaries off welfare (and more generally out of poverty), given their profile and the local context. Personalisation refers to how finely the proposed help is tuned to individual situations, as this seems to be an essential condition of successful integration. Local patterns of income support compared Helsingborg and Gothenburg The Swedish system of social protection is well known for its comprehensiveness and generosity. Even though Sweden has experienced many changes since the mid-1970s that have affected the financial basis of its welfare system (decrease in the rate of growth, higher unemployment, important influx of refugees), it has been able to maintain a high level of social protection for its citizens. Social assistance plays a small role in Sweden's social security system, which relies on universal and contributory benefits. In 1990, 53% of social security expenditure was financed by the public authorities, 45% by employer's contributions and 3% by insured persons. The universal, non-contributory citizens' pension left few older people receiving social assistance. Other social benefits, such as parental insurance and fairly generous unemployment benefits (in a country with high activity rates for both men and women), explain the very residual role of social assistance. Moreover, all other means of support need to be exhausted before claiming social assistance. The welfare system operating in Helsingborg and Gothenburg, as everywhere in Sweden - the Socialbidrag - is distinct from all others in our study due to the philosophy underlying Swedish anti-poverty policy. 96 Income support measures for the poor in European cities This rejects the idea of a minimum income in favour of the more generous concept of a reasonable standard of living'. The welfare benefit constitutes a universal right, and the maximum threshold of resources for eligibility is relatively high. As a consequence, some beneficiaries receive the welfare payment as a complement to job-earned income or other benefits that total less than the fixed maximum. It should be noted, however, that Sweden has one of the toughest means-tests of the OECD regarding the treatment of income and assets. Notably, all sources of revenue are taken into account (including savings) and easily sellable assets (like a car) have to be sold after a certain time on welfare, unless they are deemed essential to getting or keeping a job. This, together with other factors (lack of information, shame, unwillingness to put oneself under the control of social workers, sources of income judged sufficient to make ends meet during short periods of relative poverty, preference for other solutions such as family help, and so on), explain the great discrepancy between the population whose income is below the threshold and the number of people actually receiving welfare assistance (estimated in certain studies to be only 20% ofthat population). Entitlement is based on legal residence in the country. This has resulted in a dramatic increase in the proportion of legal immigrants receiving social assistance in the last decade. In 1993, foreign-born residents represented 10% of the population, but 25% of social assistance recipients. As a result of immigration, the proportion of the population receiving cash assistance increased from six to eight per cent during the 1990s. The other major group receiving social assistance is the young. In 1994, 63% of the recipients were single and mainly young people (who are entitled individually to the benefit once they reach the age of 18, even if they still live with their parents). Couples with children and lone mothers each represented 15%. The measure is marked by a strong principle of public collective solidarity, with the result that no rule of subsidiarity in relation to the family group intervenes except for the expectation that spouses and cohabiting partners support each other, and parents support their children under 18 years of age. Younger children are sometimes entitled to benefit in their own right, if it is shown that living with their parents would pose a problem. The role of voluntary agencies is also very limited. Social assistance is locally organised and, at least in part, locally financed; but the central level, with the aim of promoting geographical equality, tries to minimise local differences. At the time of our study, and until recently, this aim was pursued through recommendations issued by the 97 Social assistance dynamics in Europe National Board of Health and Welfare (Socialstyrehen). However, the legal power of these recommendations and the range of variability allowed, became controversial and new standardisation guidelines came into effect on 1 January 1999, reducing the discrepancies between municipalities in the amount granted. The monetary benefit is generous — the highest of all the cities studied for all household compositions. It is supposed not only to cover all basic needs, but also the cost of personal care (for example, insurance, washing and cleaning), recreational pursuits (for example, a daily newspaper, local travel), and social contacts. This standard amount is supplemented with the coverage of actual housing costs, provided these do not exceed a maximum level fixed by each municipality. On top of this, additional payments can be granted on an individual basis in order to respond to all reasonable requests, such as medical and dental treatment or travelling costs to search for a job. Benefits are granted for an indeterminate length of time, as long as the recipients resources remain below the fixed maximum and they are actively looking for a job (they must be registered at the employment office). However, beneficiaries seem not to be too closely surveyed on this last point, firstly because the unemployment rate is growing, secondly because social workers have large caseloads, and finally because the welfare and employment offices are not tightly coordinated2". The profoundly individualistic orientation of the Swedish system of social protection means that all beneficiaries are treated as uniformly as possible. More specifically, little or no distinction is made in the treatment of men and women, even for women with young children. The breadth of childcare public services provide implies in return that women recipients minimise their childrearing activity. Much more is expected in the way of job seeking on the part of women with young children than in the other countries studied21. The programmes designed to socially support and professionally integrate recipients are many and varied, as Sweden traditionally has very active policies in favour of training and employment. According to an OECD study (1999), Sweden was the country with the highest proportion of the GDP spent on 'active measures', as opposed to cash assistance payments. Most of these measures are implemented by the local employment offices under the aegis of the national government. Social assistance is thus limited in scope and in many cases beneficiaries switch over from welfare to another programme. The feeling expressed by many recipients was that they were only moving about within the system — back and forth between social assistance, periods of paid training, spells at Income support measures for the poor in European cities protected jobs, and unemployment benefits - raising the question of how efficient and relevant all these programmes are. However, the fact shown by our comparative analysis that, on average, people leave the programme much more quickly here than elsewhere (even if this is associated with a non-negligible proportion of re-entries), does indicate that the Swedish programmes are in comparative terms quite efficient. The relationship between beneficiaries and social assistance services is an impersonal'service-user' type. Beneficiaries have rights that the service applies in a highly bureaucratic fashion. Personal contact is limited and marked by the relational distance characterising administrative management, especially for people without social problems. Even the individualised handling of situations, whether regarding additional payments or training and employment programmes, does not significantly modify the nature of this relationship, given that the management of individual situations is highly structured with precise instructions that leave relatively little room for discretionary power. Nor does the frequent rotation of personnel facilitate a more personal relationship with beneficiaries. This relational pattern has been partly internalised by beneficiaries, who often expect little more from social workers than the financial support provided (social workers are not usually in charge of training and employment programmes) and are satisfied with an impersonal relationship. Still, many - and especially single mothers with children - complain that they do not get enough practical information or support and are not listened to attentively. There are very few differences between the two cities studied, which is an indication that, although the administration of social assistance is decentralised, the national regulations, together with a shared consensus over its meaning and underlying values, are effective in maintaining a large degree of homogeneity throughout the country. The only clear difference at the time of the study concerned the amount granted, which was higher in Helsingborg, especially if paid only for a short time. Gothenburg had already applied the new national regulations, which establish that some items previously included in the calculation of the standard amount (for example, furniture, household utensils) should now be handled on an individual basis. Labour-related programmes were until recently run by the State and not the municipalities, again tending to lessen local differences. In these times of high unemployment, however, the active role of municipalities has grown here, like elsewhere, and more significant differences might develop in the future. Social assistance dynamics in Europe Bremen and Halle The German welfare system is often identified as a typical case of an insurance-oriented system, meaning that""most" benefits depend on a contribution to a specific social insurance programme and are earnings related, aiming at maintaining the standard of living of the recipient. The system is divided into three main branches: the first gives security against old age, unemployment, invalidity, maternity and death of the main earner. This branch is based on contributions, although there are non-contributory schemes for civil servants, soldiers and the judiciary. The second branch gives compensation to victims of crimes and in cases of war injury: these programmes are non-contributory. The third branch covers social assistance and has traditionally occupied a residual position, the overall system being built around full-time employees and a male breadwinner model. Its aim is to provide a supplement to existing benefits for those who cannot contribute (or do not contribute enough) to the social insurance programmes. Until recently, after the economic boom of the postwar period, poverty was treated as both a marginal and shameful phenomenon (Paugam, 1996; Schultheis, 1996a). Although the situation has changed considerably in recent years, due notably to economic changes at world level and the reunification process, these cultural views seem to persist. This may explain why the non-take up rate in Germany is particularly high in social assistance (various studies estimate the rate at between 33 and 50 per cent). The main social assistance programme, the Sozialhi1fc,has two elements: Hilfe zum Lebensunterhalt, or general assistance, and Hilfe in besonderen Lebenslagen (assistance for people in special circumstances), directed at disabled people, people without health insurance and anyone with special needs. Pensioners whose earning-related pension does not reach the subsistence level may receive an income-related social supplement to bring it up to the social assistance threshold. Moreover, people who have exhausted unemployment compensation can claim Unemployment Assistance for a period (this is based on combined insurance and assistance criteria). Sozialhilfe is administered by local authorities and funded by these and the Länder. Anyone resident in Germany can receive social assistance, although there are specific qualifications required for immigrants. EU citizens who have been working in Germany can receive benefits for up to six months (after losing their job), but after that they have to leave the country. Similarly to Sweden, an increasing number of recipients are Income support measures for the poor in European cities immigrants, but also refugees and asylum seekers. In 1992,50% of families receiving assistance were not of German nationality. The number of recipients of social assistance has increased dramatically over the last two decades (over 150%). This rise is attributed lo unemployment, which is more acute in the former East Germany. In 1992, 5.8% of the total population was receiving social assistance, 62% of whom were women. Single householders and lone parents (mainly women) were the largest groups of recipients. However, increasing unemployment is also bringing young people into social assistance. The general welfare measure — Hilfe zum Lebensunterhalt — is similar to the Swedish one in several respects, being a national measure managed at the municipal level and thus decentralised. It constitutes a universal right to wliich access is determined on the basis of a fixed maximum amount of resources. There is a basic benefit, plus various additional amounts — namely for rent — that are determined on an individual basis, but within strict regulations. It is granted for an indeterminate length of time, as long as the recipient's resources remain below the established threshold, and applies the principle of 'something in return' by demanding that beneficiaries look actively for work or participate in training and employment programmes. Finally, the relationship between professionals and beneficiaries is first and foremost an impersonal service-user one, centred around the financial benefit. In what respect does the German system differ from that in the other countries? First, income support is conceived as subsidiary to family solidarity when it comes to access and protraction. This means that parents have financial responsibility for adult children living with them and vice versa (in some cities this also applies to non-cohabiting parents and children). Along the same lines,alimony comes before public assistance in the case of divorced persons, giving rise to bureaucratic and legal collection practices that are sometimes painful for the applicant — a fact that discourages many from applying. This 'familistic' orientation is likewise reflected in a still quite salient cultural and social model that has very deeply rooted assumptions concerning the gender division of labour and the private nature of childrearing. Public childcare services are among the least developed in Europe, and mothers of young children are treated quite differently from other beneficiaries. The German system makes special financial provisions for mothers of children under the age of three, which may be collected concurrently with the minimum income payment. Moreover, the requirements of actively looking for "work and accepting any reasonable job offer are suspended for lone parents of children under three and Social assistance dynamics in Europe applied with great leniency for single mothers with older children. In this respect, the German measure is radically different from its Swedish counterpart. The only point in which it is profoundly individualistic is in relation to equivalence scales, which are quite generous. This represents a major difference in orientation compared to other countries, which otherwise share the German familistic approach. The equivalence scales applied in Spain in cities such as Barcelona andVitoria, assume a much higher degree of sharing and saving within the family, and are therefore far less generous, reflecting concerns opposite to those in Germany (see Schultheis, 1996b). The German systein is less generous than the Swedish one not only in terms of financial support, but also in training and employment programmes, which are less dynamic than in Sweden. Furthermore, when it comes to social, as distinct from professional, integration, the principle of subsidiarity comes into play: priority is given to mutual assistance groups and other non-governmental organisations, which have a vast field of action and are partially financed by the State, thus reducing the field of overtly public intervention. Consequently, even more than in Sweden, in the eyes of its beneficiaries, public social assistance boils down primarily to its financial component. It should be emphasised that administrative agents in both Bremen and Halle have heavy caseloads. This results in limited information being provided to beneficiaries, limited availability of professionals and little time for individual attention, of which many beneficiaries complain. When comparing the two cities, one has of course to take into consideration the tremendous impact of the reunification process on the citizens of Halle. The local characteristics regarding the implementation of welfare assistance and the way it is experienced by recipients and non-recipients alike must be linked to the major structural and cultural changes that have occurred in recent years. To mention just a few aspects: before 1989 unemployment virtually did not exist and poverty was much less of a problem than today. Lone parents were stigmatised much less, and the activity rate among women was about 91%. It is now around the West German level (60%), while a number of social benefits for mothers (like very cheap public childcare centres) have been dropped. Against this background, we can observe that the actual implementation of the measure seems more favourable to beneficiaries in Bremen than in Halle. First, in the latter town the principle of subsidiarity is applied more strictly: parents and adult children who do not live together are in some cases declared financially responsible for each other, which is not the case in Bremen2-. Income support measures for the poor in European cities The second difference concerns attitudes towards the principle of 'something in return'. In Bremen, emphasis is placed on the idea of a voluntary commitment, especially in the matter of assisted employment programmes (Hilfe zur Arbeit). These are quite numerous and relatively efficient (it is estimated that two out of five recipients who participate in work schemes find a 'normal'j ob afterwards), and the rare penalties imposed for not participating are light, although after a certain time on the programme some pressure is exerted. The notions of obligation and sanction intervene far more often in Halle in order to save money and to combat a perceived tendency of recipients to settle down into being on welfare in a city where unemployment is over 20%. Last of all, beneficiaries are received differently in the two cities. While Bremen is relatively well endowed with social assistance centres, Halle has only one social service centre for an estimated 12,000 beneficiaries. This is reflected in the interviews with beneficiaries conducted in Halle. Many complained of the bad physical state of the offices, overcrowding, waiting in the halls, and lack of organisation. Rennes and Saint Etienne The French system of social protection has been built around a system combining social insurance provisions for the employed and their families, and social assistance for poor persons unable to work (invalids, disabled and older people), or otherwise exempt from the obligation to work, such as single parents with a child below the age of three (Castel and Laé, 1992, pp 10-11). Analogous to the German one in many respects, the system is possibly more complex and fragmented. In addition, benefits addressed to families with children play a much greater role in France than in any other of the six countries. Social insurance is divided into several schemes based on occupation and categories of workers. The Regime General provides coverage for sickness, work-related injuries, maternity, the family, old age and death. Specific schemes exist for farmers, civil servants, railway workers and miners. Supplementary social security schemes also include a supplementary pension. Besides these, France has a highly fragmented social assistance system, due to the existence of several programmes designed for specific circumstances. There are seven social minimum benefits including the Revenu Minimum ď Insertion (RMI), the Allocation de Parent holé, Allocation áe Solidarite Spécifique, Minimum Vieillesse, Allocation Veuvage, Minimum Invalidite and Allocation auxAdultes Handkapés. Whereas the RJVII is targeted at those not eligible for any other scheme, the other Social assistance dynamics in Europe six are targeted at specific populations. There is also a complex interaction between the various social programmes. In the French context, the fragmentation of programmes requires a closer look at each one. Three programmes that are outside the scope of our study (Minimum Vieittessc, Allocation aux Adultes Handicapés (MAH) and Minimum Invalidite (MI)) cover respectively: the population over the age of 65 (or 60 if disabled), people with disabilities, and people who are sick or disabled through a non-work related accident. These programmes include resident immigrants. However, whereas to receive the Minimum Vieilkssc a person needs to have 10 years residence, for the MI no specific length of time is required. Two family-related programmes are the Allocation de Parent holé (API) and Allocation Veuvage (AV). The first is a benefit for lone parents and the second for widows less than 55 years old. The API is more generous than the RMI, but it has a time limit (or age of child limit) and the claimant cannot reside with a member of the opposite sex. The AV is for widows who are rearing or have reared children. It excludes the self-employed (only farmers are included) and is financed through social security contributions. While these two programmes are means-tested, there is also a complex system of universal benefits paid for children, which is administered by the Caisse d'Allocations Familiales, and represents a substantial integration to the income of families with children. One other programme, Allocation de Solidarite Spécifique, is a state benefit for the unemployed whose entitlement to unemployment insurance benefits has expired. Beneficiaries must be under the age of 65 and have worked for five years out of the previous ten. This programme is mainly directed at long-term unemployed people. It is financed two thirds by the State and one third by contributions from civil servants. One of the principal features of welfare provision in recent decades is the blurring of boundaries between social insurance and social assistance. This can be seen for instance in the gradual widening of social security benefits to include the unemployed as well as people who are too poor to make financial contributions. At the same time, non-contributory and means-tested social benefits, such as the RMI, are being introduced, marking a radical departure from the insurance principle. As new forms of poverty developed during the 1980s due to unemployment and social precariousness, the distinction between the deserving and the undeserving poor gradually lost its relevance. At the same time, it became clear that an attempt to divide poverty into different categories made little sense when they consisted of an increasingly heterogeneous population. It was in this context that France set up a 104 Income support measures for the poor in European cities universal minimum income system (the RMI, Revenu Minimum ď Insertion) in 1988. The guiding principle is that every individual whose income is below an objective maximum level has a right to the RMI, regardless of the reasons for their poverty. The RMI, however, was not designed to replace the existing social minima that had developed over the years to cover specific types of situations. Its aim was more to 'fill in the holes' of existing welfare provision. The RMI thus has a somewhat residual role, as it is subsidiary to a host of category-based measures that already make up a form of minimum income for many people (and indeed are often more generous than the RMI). The French social system emphasises the principle of individual entitlement and national responsibility, over family membership and family solidarity. Yet, there is an age limit: applicants must be over 25, unless they have children or are pregnant. The main reasons given for this minimum age are the preservation of family solidarity, the danger of 'disincentiving' young people from working, and the existence of a policy of social and professional integration aimed especially at the 18 to 25 age group. Another reason cited is the cost that would be involved in extending the RMI to all persons over the age of 18. Yet, the 18 to 25 age group is undergoing serious difficulties, and the question of extending the RMI to cover them is regularly discussed. Immigrants are eligible, provided that they have been resident in France for three years. The RMI is a national measure whose financial characteristics for the most part do not vary from city to city. In contrast to the Swedish and German cases, no portion of the financial benefit may be individualised. It varies only according to household composition and resources. Only the added housing benefit varies: according to the level of the rent and up to a maximum amount. However, since this maximum is rather low and therefore frequently reached, one can hardly speak of individualised treatment in this case either. The RMI is granted for an indeterminate length of time. It involves a notion of mutual commitment, rather than 'something in return'— therein lies its originality. Recipients commit themselves to participating in activities that they help to define, aimed at improving their situation. The agreement is formalised by the signing of an integration contract, which implies that in return the community will develop suitable proposals for integrating the beneficiary. The integration contract involves no firm obligation, because its objective is not so much to constrain the recipient as to support people as they try to achieve their projects. Penalties are applied only in cases of manifest bad faith. Integration is either social (for example, regarding housing, health, hygiene, treatment of drug or 105 Social assistance dynamics in Europe alcohol addiction) or professional, depending on whether or not tile recipient is fit to work. Contracts are signed for periods between three months and one year and may be renewed as often as deemed relevant. The focus on integration activities, moreover, implies institutional cooperation among different local actors, (public, non-profit, private; for example, enterprises), who are responsible both for providing resources for the integration contracts (temporary in-work experiences, training, and so on) and for implementing them. The fact that there is an institutionalised connection between financial benefit and integration activities plays an essential role in the relationship established between professionals and beneficiaries. While the delivery of benefit is handled by bureaucratic procedures, as in Sweden and Germany, this is not in theory the core of the beneficiary's relation with the social service professionals, which is focused on integration activities. The personnel in charge of receiving beneficiaries are not the same as those who manage the delivery of the income allowance. Their job is to inform beneficiaries of their rights and to assist them in administrative procedures, as well as to meet them regularly to see how the integration activities are proceeding. This facilitates personalised follow-up. The recipient's perception of this relationship is conditioned therefore by how attentively their case is actually followed and how useful the proposals and projects for achieving integration are. Issues of access and financial benefit are managed by a national organisation and there is no margin for local action. Above and beyond local demographic and socioeconomic conditions then, any variation with regard to this overall framework has to do with integration policy. The only other possible source of variation concerns additional forms of support, either financial or in kind, which may be attributed by municipal or departmental social services. In Rennes, for example, welfare beneficiaries may use public transportation free of charge, whereas this is not the case in Saint Etienne. This was the only difference we were able to observe at the level of income support. Integration policies, on the other hand, differ considerably from one municipality to the next. Though orientation and funding are fixed in a national regulatory framework, such policies are essentially applied at the local level. Modes of social intervention in Rennes, for example, are much more dynamic than in Saint Etienne. The city of Rennes has set up special personnel to receive welfare beneficiaries and has specially trained local integration workers (as distinct from social workers), who assist beneficiaries in all matters of professional integration. Moreover, the high level of inter-institutional coordination in Rennes has allowed Income support measures for the poor in European cities tor the development of quite diversified offers in the domain of professional integration, whereas Saint Etienne has been characterised for many years by institutional compartmentalisation and much less active recipient follow-up (although the situation was improving at the time of our study). The interviews conducted confirm that follow-up in Rennes is more attentive, personalised and efficient than in Saint Etienne. Overall, Rennes is one of the most dynamic of the cities studied. Barcelona and Vitoria The Spanish welfare system seems to present a strange combination of precariousness and relatively limited exclusion. In fact, despite a high proportion of households (between 15 and 25 per cent) living below the poverty line - defined as 50% of the average equivalent family income (estimated during the 1980s and early 1990s) - and despite high unemployment rates (about 20% in the mid-1980s, and again since 1992), Spanish society seems to display a relatively limited degree of social exclusion. There are few homeless people, lower and decreasing crime rates than in other European countries, low infant mortality, and low rates of single person and lone-parent households (Laparra and Aguilar, 1997). The presence of high integration levels depends on various factors, including family solidarity23, that make it possible to reduce the impact of unemployment and job instability (according to official statistics, one third of the labour force had a non-permanent job at the time of our study), which encourages the various household members to adopt combined strategies. The Spanish social protection system, as in other Southern European countries, offers greater protection to older people (aged over 65), people with disabilities (invalidity greater than 65%) and, in general, to those more unable to work than the able-bodied poor. At the same time, among the able-bodied, it protects those who have suffered from precariousness in their working history (insufficient contributions during their working life), more than those who for some reason have been excluded altogether from the labour market (persons who have never succeeded in finding a job). Income support measures for the poor - Ingresos Miniums de Insertion, Renta Minima de Insertion or Salario Social - have developed at the regional level and the laws that regulate this kind of measure vary greatly in the 17 Communitades Antonomas (Autonomous Communities)2"1. Since there has been a high discretionary power in implementing the national guidelines, these programmes have not been developed in the same way or to the Social assistance dynamics in Europe same degree across the country. The main differences concern the fact that in some regions benefits can be legally claimed as a right, while in most cases the regional governments grant the benefit at their own discretion (Laparra and Agüilar Hendrickson, 1997, p 528). In five regions (Andalusia, Asturias, the Canaries, Castilla-Leon and the Basque country), there are workfare programmes with an obligation to accept job offers. These measures aim to encourage labour market insertion, and involve the obligation to sign a social insertion contract. They are very different from one region to another, and in some cases claimants may be asked to work within social services assisting older people and/or people with disabilities. Six regions have legally established temporary social interest jobs (Emplejo Social Protegido) as a compulsory alternative to benefits25. The Basque25 and Catalan systems — the Plan Integral de India contra la Pobreza, which includes the IMI (Ingreso Minima de Insercion), the Ayudas ilc Emergenci a Social (AES) and other accompanying measures inVitoria, the PIRM.I or Programa Interdcpartamental de Rentas Minimas de Insercion, which includes RMI (Renta Minima de Incercion) as well as emergency and accompanying measures in Barcelona — share the same cultural inspiration. Claiming benefit is considered a right and all those who are entitled should obtain it. The reference unit for this kind of support is the household (which must have been formed at least one year before applying tor benefit). Benefits are given to persons between 25 and 65 years old. As in France, it can be given to younger households if there are children or people with disabilities needing care. The basic amount granted is identical to the monetary level required for admission to the benefit and is quite low (lower in Barcelona than in Vitoria). Moreover, in contrast to Germany, the equivalence scales are rather unfavourable for large families. InVitoria, however, the basic amount is systematically supplemented with an additional sum granted on an individual basis, with the aim of covering part of the rent and current housing expenses. In Barcelona there is no additional amount, but at the beginning of the period of assistance, social workers can give money to pay debts incurred up to that moment. In this respect the cities are vastly different: whereas in Barcelona the basic amount is usually not enough to live on, inVitoria it is generally sufficient to stabilise living conditions at a minimum level. It should be noted, however, that family solidarity and the informal economy (which is much more developed in Barcelona) enable many beneficiaries to pull together different types of resources. In both cities the benefit is designed for one year, but it can be renewed as long as the condition of need persists27 and as long as the beneficiaries Income support measures for the poor in European cities meet the criteria for attribution, which include a 'something in return' obligation, most often in the form of a signed integration contract. The main difference with the French RMI concerns the compulsory participation in integration activities. Evaluating the situation created by this land of benefit, concern that financial support given to the poor might discourage the search for a real job seems unfounded. The many programmes set in place to help beneficiaries to get out of welfare, together with the low amounts granted, render it difficult to argue that it creates a culture of dependency. It should be added that, at least among the beneficiaries interviewed for this study, being on the measure was considered by recipients more a privilege than a stigma. With regard to the accompanying measures2", there are courses in elementary education, professional training, reorientation and refresher courses. The same kind of courses are offered to most beneficiaries: they have not proved diverse or differentiated enough to suit recipients'profiles. Such courses are usually more appropriate for the most deprived recipients — among them immigrants29— namely because they ensure re-socialisation and improve their self-confidence; but are less suitable for people whose main problem is lack of an adequately paid job. In addition to these courses, the municipality ofVitoria has reserved many low-qualified temporary jobs, such as cleaning or concierge work, for welfare beneficiaries. It can happen that people get off the programme because they obtain such a job, keep it long enough to be able to collect unemployment insurance, then when they exhaust this, return to social assistance if they do not find anything else. In Barcelona, the job offer is much less institutionalised or institutionally controlled than in Vitoria, and getting a job has more to do with the social workers'personal initiative. The relationship of recipients with social services and social workers differs greatly in the two cities. InVitoria a rigorous surveillance system has been put in place, with the aim of checking recipients' resources, prevent fraud and ensure that the obligation to look for work is respected. This has a direct effect on social workers'practice: although they regularly meet beneficiaries, they must fulfil numerous bureaucratic procedures and are consequently often not available to follow-up beneficiaries, either to give them moral or psychological support or integration assistance. Beneficiaries are often critical of the service and personnel, though some declared that the pressure social workers exerted on them ultimately had a positive effect. In Barcelona, on the contrary, social workers tend to have an extremely understanding attitude towards beneficiaries, giving them much advice Social assistance dynamics in Europe on job openings in both the formal and the informal economy. They are not strict about the 'something in return' obligation, or about declaring income sources from which the recipient is known to benefit, as long as they do not exceed a certain amount. Such treatment expresses the social workers' intention to practice a kind of local justice in the face of a measure they judge insufficient — an attitude that understandably wins a very favourable judgement from beneficiaries, who are otherwise quite critical of the financial sum allotted and the kind of professional integration projects proposed. The other side of the coin is that this situation can lead to high discretionary power, collusion and consequently mistrust and manoeuvring. Milan,Turin and Cosenza The social security system in Italy is highly diversified and follows a corporatist model. Schemes are earnings-related and separately funded. However, there is also a Public National Health System, established in 1978, covering all citizens and legal residents for medical care. Old age pensions are contributory People whose contributory record has reached a minimum period, but is not sufficient to qualify for the minimum pension, may have it integrated on the basis of a means-test for the individual or the couple (Pensione Integrata al Miniino). A Pensione Sociale provides a means-tested pension for people over the age of 65 who do not benefit from any of the other pension schemes3". Invalidity pensions and invalidity allowances are of two kinds: one is provided to those who have become disabled during their working life; the other is for those who were born disabled or became so while not in the labour market. The latter (Assegno di Invalidita Civile) is means-tested. Another item in the social security system are the different kinds of unemployment benefit. There are two main types of unemployment benefit. One is granted to workers made redundant due to a decline in business (Indennita di Mobilita) and the other to involuntary unemployment in the case of individual dismissal. The first is much more generous (about 80% of the previous wage) than the second (30% of the previous salary). Young unemployed people, and people of any age (mostly women) without prior work experience, are not entitled to unemployment benefit or any other type of benefit within the social security system. Finally, there is no universal system of child allowances. Means-tested family allowances (Assegno al Nucleo Familiäre) are paid only to wage workers, since these benefits are partly financed through contributions31. Income support measures for the poor in European cities There is no nationally regulated means-tested income support, except for older people (Pensione Sociale) and people with disabilities (Assegno di Invalidita Civile). Since 1977, the State has transferred responsibility for social assistance to the regions and the municipalities. In the absence of a national law regulating rights and duties in social assistance (and social services provision)32, many regions introduced regional laws on social assistance. Not all regions, however, have such a law. Moreover, there is diversity within regions as well as between regions in the access criteria to social benefits. The consequence of this is a highly locally differentiated system of social citizenship. Eligibility requirements for means-tested income support (with the exception of those nationally regulated) are defined locally and mostly defined in categorical terms. In general, they privilege older people and people with disabilities and children, in that order33. Conditions of eligibility are more restricted for the able-bodied than for older or severely disabled people. The able-bodied need to be ready to take a job if offered or accept jobs of public utility. As in Germany, benefit is not officially payable when there are relatives legally liable for any member in the claimant household. Unlike Germany, the range of'obliged kin' is quite wide, including brothers and sisters, parents and children in law, uncles/ aunts and nieces/nephews. Plowever, in real life this requirement may be overlooked. The amounts paid vary substantially among regions and among cities. Some groups receive more generous benefits, as is the case for pensioners and disabled people, whereas the able-bodied (including lone mothers) are less generously treated and often totally excluded from any income support measure. The great heterogeneity of the support measures available (in some municipalities there are none at all), and the even greater heterogeneity of their modes of implementation, underlines the need to be extremely careful in analysing differences and similarities among different cities. For example, if we consider the minimum living allowance, Minimo Vitale (MV), which is provided in the three cities examined, Turin, Milan and Cosenza, we find three very different situations. In Milan and Turin, this allowance is paid to all citizens who do not reach a minimum income threshold considered essential to live on. In Milan, it is conceived in principle as a social right, but is implemented with a high level of discretion. In Turin, the discretionary power is low, even though from a legal point of view receiving income support is not considered a right, and people denied the benefit cannot claim for it before a court. In Cosenza, the MV is a category-based two year measure reserved for single mothers that, due to its limited impact (given the small number of lone mothers) and atypical nature, Social assistance dynamics in Europe cannot be compared with the allowances in Turin and Milam A more general measure, though as we shall see quite limited, which is applied in Cosenza is the Assistenza Economica. In order to be able to compare broadly similar measures, we have looked at the MV and basic income for food needs (Minimo Alimentäre, MA) in both Turin and Milan, and the Assistenza Economica in Cosenza. Generally speaking, the minimum living allowance (MV) is equal to the minimum contributory social insurance pension level and there are local equivalence scales to take account of different household sizes. In Milan, however, there is no clear connection between the amount and household composition. The amount paid and entitlements vary greatly between Milan andTurin,but there tends to be a basic distinction between the treatment of older or disabled people — who usually receive the amount necessary to bring them up to the minimum social security pension level - and the treatment of able-bodied applicants, which is much more discretionary. Different groups receive different levels of benefit. For example, pregnant women and lone parents may have their income made up to the social pension level with a supplement for rent, although for a limited period. For other groups of recipients, income support is less generous. In general, the benefit is higher in Turin than in Milan (where it is even lower than in Barcelona). Moreover, in Milan34, those who fulfil means-tested conditions are not automatically entitled to a monetary subsidy and their £right'in fact depends on the constraints of the city social welfare budget. This is not the case in Turin, where discretionary power tends to be exercised through a more rigorous control of resources and a prolonged waiting time for the renewal of benefit for applicants deemed undeserving. Discretionary power may also affect which of the two benefits is granted (the MV is for food, clothing, general and social life expenses, while the MA only covers expenses related to food, and is therefore less generous). Such power means that the first meeting with the social worker is crucial and that the applicant's negotiating skills can make a difference. Furthermore, in both cities family solidarity is given primacy, increasing selectivity for admission to either of the measures. In Turin the full amount of MV is paid to able-bodied claimants for a maximum of one year. If their need persists after that, they may continue to receive support, but at a lower rate (MA instead of MV). In Milan, the period of payment is in practice always limited in time (3-6 or 12 months), but again there is a disparity among different welfare offices catering to different categories of beneficiaries. While the assistance provided for adults in need (the UAD service) offers short-term allowances, that for Income support measures for the poor in European cities lone mothers (the SSMI service) offers medium-term allowances, and that for older people (CADA service) long-term payments. In both cities, the amount received by able-bodied recipients is generally insufficient to live on and must be supplemented with income from the informal economy, family assistance, or private help from charities. Neither MV nor MA are conditional on taking job offers or participation in public work projects; but able-bodied recipients must demonstrate their availability for work, which usually means that applicants have to register with the unemployment office. In both Turin and Milan some kind of temporary job for people 'at risk of marginality' is provided: for example, the so-called Börse Lavoro (job grants), which last for a period of six months and subsidise employers, or usage support, a financial contribution to firms that employ young people in need indicated by the social services, or introductory apprenticeship, work experience for young people in need. In Turin there is also a wide recourse to Cantieri d i Lavoro and socially useful jobs, which provide publicly funded temporary work in the public and private sector (these are protected jobs for unemployed people that help connect them to the labour market with activities of social utility, under laws 55/84 and 608/1997). The situation in Cosenza is totally different from the above. The economic assistance for the poor (Assistenza Economica) does not provide sufficient financial resources even for the most elementary living expenses: the yearly amount is similar to the lowest monthly amount in Turin and Milan (about Ecul20). The inefficiency of the measure is further compounded by a series of administrative and organisational deficiencies. Notwithstanding the negligible amount, three levels of need (and of benefits) are distinguished in the regulations. Access to the first amount — designated for ordinary situations — is granted automatically to all those whose revenues fall below a certain level. The two other types (particular cases and exceptional situations) are specified in the regulations, but their content is not formally defined. Criteria to distinguish 'particular' situations have been established informally by the municipal social service in its daily work. In this poverty-stricken context, where unemployment is above 30%, beneficiaries are provided with no training or job-search guidance and nothing is demanded of them. Relations with social workers are generally very poor, and characterised by laborious bureaucratic procedures and checks that beneficiaries deem humiliating and offensive, given the absurdly low benefit paid. Moreover, many respondents said they had witnessed preferential treatment that had nothing to do with applicants' real needs. They often feel that social workers have no empathy with them and do 113 Social assistance dynamics in Europe not give them accurate information about the benefits they could obtain. At the same time, some of them make an effort to use the informal negotiation process to their favour, to obtain additional advantages. Lisbon and Porto The Portuguese social security system is very recent in comparison with the vast majority of European social systems. Only in 1984, with the publishing of the Basic Law of Social Security (Lei de Bases da Seguranga Social 28/84) did the principles upon which the new system would develop become clear. Tins basic law represents an enormous simplifying effort, reducing the numerous former regimes to two, enhancing the administrative rationalisation and improving the visibility of rights and duties of beneficiaries and employers, not to mention the State. The law is based on the idea of the 'universal right to social security' and the main aim is the protection of workers and their families, and the protection of persons suffering from the lack or loss of means of subsistence. Since there were still very few measures covering people of working age, a new non-categorical benefit35 — the National Minimum Income — was added to the non-contributory system in 1996. At the time of our study, it had been operational only in some municipalities for a short time. Here we shall examine the measures operative in Lisbon — Subsidio Mensal — and Porto — Prestagöes Pecuriiárias de Acgäo Social — before the application of the new minimum income programme, which is at present going through a trial period and for which we lack empirical data at the local level. Neither measure can be considered as a right, since entitlements depend on subjective evaluation by social workers and on resources of the welfare services. In Lisbon social assistance (even after the introduction of the new measure) is run by a religious organisation, the Santa Casa da Misericordia de Lisboa (SCML), which although state-regulated, has its own funding sources and enjoys a great deal of autonomy of action. In Porto social assistance is managed by the Regional Centre for Social Action, a public structure that also has a high degree of autonomy. In both cases the measures are in theory applied to all persons living in poverty, but in practice they are highly selective and category-based, though the categories are not well defined. In practice no maximum level of resources is taken into account in deciding who is admitted to the programme. Even though in Lisbon the SCML has defined a household income threshold, this is usually totally disregarded and the decision is essentially based on how the social worker Income support measures for the poor in European cities evaluates the situation. The criteria on which judgement is based call into play a great deal of unregulated discretionary power. The factors considered may include the applicant s personal resources, basic expenses (rent and associated expenses), help available from other sources such as family, friends, neighbours, and charity organisations, the urgency of a given case, the social worker's empathy for the person and how they judge the persons attitude, and the funding actually available. In both cities, the amount given is generally low and constitutes only a portion of a person's overall resources. There is no precise definition of the monthly amount and there is no clear relation between household composition and amount granted. Benefits are granted without a maximum duration, yet there are wide differences in the implementation. Single mothers with children, and children in general, have priority. In Lisbon limits on duration seem less strictly enforced than in Porto, where the city's resources are much more limited. This difference also shows up in the average length of time on welfare, which cannot be explained by differences in the structure of opportunities in the two cities — over 30 months in Lisbon, and much shorter in Porto (see Chapter Five). Action to promote the finding of new employment is left primarily to the social worker s personal initiative. The limited resources do not permit an efficient monitoring beyond giving information on training programmes and job possibilities. It should be underlined, moreover, that due to the high degree of selectivity and the extremely negative profile of beneficiaries and their family entourage, few beneficiaries are actually in a position to hold an ordinary full-time job. When people go off the programme, it is rarely because the problem of poverty has been resolved, and there is a high risk that those who do leave welfare will come back onto it later. Beyond the social assistance measures, other initiatives have been introduced in the last few years at the local level in order to combat social exclusion. Among these the Local Initiatives of Development and Job Creation should be mentioned. These ami to create jobs by supporting the creation of small businesses. Employment policies are partly covered by initiatives organised at the national level (but locally implemented) in the context of the Prograinas operacioiiais do IEFP and partly at the local level (Iniciativas local do Emprego). Two studies performed in the city of Lisbon (Cardoso, 1993; Silva et al, 1989) indicated that a significant number of demands for social assistance had no response from the existing institutions. Some institutions face serious financial constraints, which means that only the demands of the most needy are satisfied. Moreover, in some cases the demands are Social assistance dynamics in Europe addressed to the wrong institutions. Given the fragmentation of institutions, each with its own 'public', it is not always easy for beneficiaries to find out which one they should address. Cardoso in particular points out that a significant percentage of poor households in Lisbon (52% of the population investigated) do not ask for social assistance, either because they do not know of the existence of the services (42%) or they have a bad opinion about their efficacy (32% of cases). Another reason is the Stigmatisation effect of social assistance, which leads some people to refuse the assistance they might be entitled to. According to our research, recipients' relations with social services are characterised by dependence, strongly coloured with fatalism and subjectivism (but not patronage or 'clientelism', as in Cosenza). Against this common background, however, there are differences between the two cities. In Porto relations are strongly affected by the extreme selectivity with which applications are considered, by the utter opacity of decision criteria and the absence of explanations for decisions, and finally by a caseload that precludes any personal follow-up. Many of those interviewed said they had stopped going to the service centre, either because their requests for assistance had been repeatedly rejected or because they had been taken off the programme without explanation. Evaluation of both services and social workers was generally quite negative. In Lisbon, on the other hand, we observed a dissociation in people's minds between social services as suchjudged negatively due to the laborious administrative procedures, lack of information, and so on — and social workers as people, who are often judged positively for their monitoring activity, support and dedication in making use of available funds. The following figures constitute an attempt at graphically summarising this analysis of the different income support systems, in all their variety and complexity. They should also offer a visual map for better interpreting the results of the study on social assistance careers presented in the next two chapters. Income support measures for the poor in European cities Figure 3. S: Swedish cities Bureaucratic regulation versus discretionary power Activation measures Duration D I I I I I I I I Generosity Family obligation versus f~ collective solidarity i i , , , i , i—, Universal versus category- based orientation Selectivity Recipients' duties [| Gothenburg and Helsingborg Figure 3.2: French cities Bureaucratic regulation versus discretionary power I0x Activation measures Duration Generosity Family obligation versus collective solidarity +_i_. Universal versus category-based orientation Selectivity Recipients' duties □ Rennes -e— Saint Etienne 117 Socio/ assistance dynamics in Europe Figure 3.3: Italian cities Bureaucratic regulation versus discretionary power 10-r Activation measures Duration I- I I I I D Generosity Family obligation versus collective solidarity 1 I 1 1 I I Universal versus category-based orientation Selectivity Recipients' duties -Eh Turin -©- Milan .....o- Cosenza Figure 3.4: Spanish cities Bureaucratic regulation versus discretionary power Activation measures Duration Generosity Family obligation versus collective solidarity Universal versus category-based orientation Selectivity Recipients' duties -y=- Barcelona -«— Vitoria Income support measures for the poor in European cities Figure 3.S: German cities Bureaucratic regulation versus discretionary power lo-r Activation measures Duration D Generosity Family obligation versus collective solidarity Universal versus category-based orientation Selectivity Recipients' duties E] Bremen -9- Halle Figure 3.6: Portuguese cities Bureaucratic regulation versus discretionary power Activation measures Duration l-Q Generosity Family obligation versus collective solidarity ,J__L_J__l_il Universal versus category-based orientation Selectivity Recipients' duties E3 Lisbon -*- Porto Social assistance dynamics in Europe Notes 1 The concept of referential' is frequently used in public policy analysis in France. See, notably, Müller and Surel, 1998. 2 Primary solidarity refers to the form of solidarity connected with family or friendship, as well as neighbourhood relations, and implies a personal acquaintance. Secondary solidarity, on the other hand, refers to more abstract social relations, inasmuch as they apply to 'anybody' within a given universe of reference (for instance, the city, the country). They cover both private forms of solidarity, like those implemented by religious associations, and public forms, whether at the State, regional or local level. 3 It is important to underline that in the ESOPO study this homogeneity is also due to the choice of the cities. Thus, the choice of Turin and Milan for Italy, and Barcelona andVitoria for Spain, brings to the fore important similarities. It would, however, be a mistake to extrapolate from this the existence of a clear-cut Italian or Spanish welfare state. For Italy, the counter-example of Cosenza is fundamental in this regard. A similar contrast would have emerged in Spain had we included in our sample a city from the south, for example, from Andalusia. 4 Even those countries where integration is not at the heart of their income support programmes are experiencing a move in this direction. Thus, in Germany, a 1993 law requires that local administrations develop labour opportunities for those out of work, notably the young, in the form of general interest jobs or fimded labour contracts. For an international comparison on the link between welfare and work, see Commission of European Communities, 1997; European Foundation for the Improvement of Living and Working Conditions, 1999; Gough etal, 1997. 5 In France, it is estimated that the ratio for access to public employment programmes between the 10 most dynamic departments and the 10 least dynamic ones is 3 to 1. See Charbonneau and Landais, 1996. ''We resorted to the 'vignettes' method (see for example, Bradshaw et al, 1993; Finch, 1987; Marradi, 1996). More specifically, we presented a number of social workers in each city with possible 'cases' of need and asked them what was available in their city in principle, and what would occur in practice. 120 Income support measures for the poor in European cities 7 Twenty-four interviews were conducted in each city in 1997, half with former and half with current claimants. For the French cities, in addition to these in-depth interviews, we also carried out a general survey on beneficiaries. s Our approach draws its inspiration from Franz Schultheis' text: "La famille, une catégorie du droit social?: une comparaison franco-allemande", in Comparer les systéines de protection sociale en Europe, vol 2, Paris, MIPJE, 1996, pp 203-34. u As with Figure 2.3 in Chapter Two, the proposed scores for each dimension have been agreed with local researchers, on the basis of local rules and patterns of implementation, with the validation of social workers within each local context. Scores from 7 to 10 indicate a high level for that specific dimension, 6 to 4 indicate a medium level and 3 to 2 indicate a low level. Score 1 means that the dimension is absent or irrelevant. '"This does not mean that the Swedish cities have absolute uniformity in decision making, only that in comparison with the other local contexts considered they can be positioned at the left side of the continuum (for intra-country diversity in Sweden, see Gustafsson et al, 1993). 11 In the city ofTurin, the possibility of renewing benefit for the entire period of need for the able-bodied poor is established not in the general regulation approved by the municipality government, but in a circolare — an interpretative document -from a lower legal administrative level. This means that social workers have considerable leeway in interpreting the regulation (see Bosco et al, 1999). 12 This explains why the use of the term 'right' in the interviews both with social workers and with beneficiaries may be equivocal. It may be meant in the sense of deservedness, that is, to have a right to assistance given a particular itinerary and situation. This must not be confused with the legal meaning of having one's rights recognised and getting what one is entitled to. 13 Except in Porto, where at the time of the study there was no income threshold and a subjective evaluation of need had to be made by social workers. 14 For example, in Tur in property ownership (except when the accommodation is suitable for the needs of the household and inhabited by it) or registered assets (for example, motor vehicles not used for work or required for health reasons) are still considered grounds for ineligibility. 121 - FOUR Maiisol Garcia and Yuri Kazepov A medium range approach Are some categories of people more likely to be on benefits because they are really more needy? Or is it because the entitlement structure privileges them as a group? It is not easy to answer these questions, as it requires a large amount of information and the consideration of many aspects. In order to give a tentative answer, we need to pull together the issues discussed in Chapters Two and Three of this book: the way in which the local context produces vulnerability and poverty, and the way in which the problem is conceived and tackled at the national and local level. We intend to examine in particular the role of the institutional mechanisms: how they work and how they influence the process of becoming a beneficiary of social assistance. In order to understand this process, we have to consider the main dimensions that influence the experience of poverty and downward social mobility, which are: (a) the socioeconomic context (including demographic changes), (b) the social network of the potential claimant, and (c) his or her personal capabilities. These dimensions, together with the institutional design, interact with one another to define the degrees of social vulnerability and the risk of becoming poor (and hence the risk of being on social assistance). Of course, the same interaction also defines the set of resources and opportunities that people may use to escape poverty. This chapter, however, focuses on the paths into social assistance and highlights how the contextual driving factors (described in Chapter Two) are transformed by the institutional setting (described in Chapter Three) into socially defined conditions of need of income support. Social assistance dynamics in Europe Figure 4. hThe dimensions influencing downward mobility Main socio-demographic characteristics of the population II. Main socio-demographic characteristics of the poor Main socio-demographic characteristics of the assisted Figure 4.1 shows how the main dimensions interact with one another. We are aware that at this stage the resulting picture is approximate and does not take into account the full range of complex interactions between these dimensions. Nevertheless, it allows us to appreciate that the vulnerable groups 'produced' by the interaction between a given population and its socioeconomic context are not necessarily in a condition of economic need per se. This need emerges through a further selection process involving the institutional dimension, which is at the core of our explanation. The institutional dimension structures people's life course by defining how public (and partly also private) resources are - or should be - allocated in relation to the risks emerging during critical status passages in life2. Such passages occur in relation to age-specific phases and specific events (entering the labour market, becoming unemployed, retiring, and so on). The link between the two is self-evident, but it is important to keep them analytically separate, as social policies address them differently. The ways in which risks are distributed institutionally among social groups become more evident by contrasting different contexts. Unemployed people deal with their vulnerable situation in relation to the labour market in different ways according to the policies that exist in the specific country; their generosity, coverage, accompanying measures, and so forth (Gallie and Paugam, 2000). It is very different having to rely on unemployment benefit of 30% of the last net wage for 180 days (as in Italy)3 and a benefit of 60-70% for nearly two years with several funded Why some people are more likely to be on social assistance than others retraining options (as in many other European countries). Of course, specific social-demographic factors, like an unbalanced age structure, can produce more (or fewer) people who find themselves in need of care and support (Micheli, 1997). But how many of these turn into cases of economic need is a matter of institutional intervention and policy design (Goodin, 1996). This implies that, on the one hand, the dimensions identified above produce social groups that are more (or less) vulnerable and, on the other hand, that vulnerability turns into concrete need because social policies, by structuring the flow of resources — both monetary and in ldnd — redistribute the risks of becoming needy. In other words, living conditions in general, and conditions of economic need in particular, are largely the result of the filtering process occurring at the institutional level. Therefore, different institutional arrangements and configurations produce large variations in the distribution of living conditions. The literature is not always clear about this process. With few exceptions (for example, Behrens and Voges, 1996; Leisering and Leibfried, 1999), most scholars (for example, Esping-Andersen, 1990, 1999; Ferrera, 1998) tend not to emphasise the relations between the different policies and the related synergy effect in relation to poverty, even though they acknowledge the influence of the welfare state in structuring the social stratification of society. The interaction among policies and among the dimensions that structure the distribution of risks in society vary according to history and geography, but at a different pace. An increasingly wide range of events can trigger downward mobility by breaking down the equilibrium between available resources and needs. The individualisation of need (Andreß and Schulte, 1998) also implies the individualisation of risks (Beck, 1992). This means that the readjustment of policies is lagging behind social transformations and that their adequacy suffers the burden of de-contextualised solutions. The strategic importance of schemes designed as the last safety net or last resort against poverty, like social assistance, is increasing. The number of claimants is rising (Atkinson, 1998b; Eardley et al, 1996; Mingione, 1996; OECD, 1998b, 1999) and the overall distribution of social expenditure within welfare policies is increasingly shifting people towards social assistance. But the adjustment of the schemes themselves to the new, emerging needs is slow. Their institutional design is often still focused on traditional forms of marginality, which in fact characterise fewer and fewer of the social assistance claimants. There are other dimensions that play a major role in the process of becoming poor and socially assisted. Taking further the analysis developed in the previous two chapters, we identify (Figure 4.1) a relational filter, Social assistance dynamics in Europe including individual competences4 and the primary social support network, which activates reciprocity resources (Cochran et al, 1990; Gullestad and Segalen, 1997). The two dimensions can play very different roles, depending on the socioeconomic context and institutional arrangements existing in a given country at the specific local level (region, city). We can identify three main patterns (Kazepov, 1999). In the first, the primary social network presents a complementary role, as in welfare systems where family and kin support is integrated into the institutional design of social assistance measures (for example, in Germany, France and most Continental European countries). In these systems, the State not only identifies the family as the main source of support, but also assists it with relatively generous transfers and/or services. This is a form of active subsidiarity. The second pattern is more substitutive and occurs mainly in countries where little or no direct institutional support is given to families (for example, in Southern European countries). Here the family is burdened with many social responsibilities and has to pool and share resources more than elsewhere in order to cope with events which could trigger downward social mobility. This is a form of passive subsidiarity that presents problems of sustainability in periods of continuous transformation and hence increased vulnerability, which redefine the conditions of social inclusion and exclusion. The third pattern is characterised by an additional role played by reciprocity networks in welfare systems where the State intervenes extensively for all needs that may arise in the population. Here, the family support network is not burdened with social responsibilities, and the individual is the bearer of substantial rights. Family intervention adds resources to the assistance already granted institutionally. There are different mixes of these three patterns even within single countries, especially where the socioeconomic context and institutional setting vary considerably at the regional level, like in Spain and Italy. Generally speaking, a sort offunctional equilibrium develops between the form of family support existing in a given context and the institutional design of income support measures. However, the different role assigned to family solidarity affects the outcomes of public intervention in terms of distribution of income inequalities. It is not surprising that; the most unequal distribution is to be found in systems where the family plays a strong role, while the least uneven is found where the family plays an additional role5. This outcome is the result of a complex interactive process that highlights the difficulty in breaking down the reproduction of inequalities, particularly for vulnerable families and groups, without the intervention of external resources. In substitutive contexts the family, nn Why some people are more likely to be on social assistance than others whose role is also the result of a specific ideological and cultural background, is forced to cooperate and pool resources. In addition to the socioeconomic and demographic filter, individual competencies and social networks also influence the access to resources and therefore the risk of becoming poor (assisted or not). In order to understand this influence, however, we have to consider it together with the institutional dimension, which determines the family's support strategies. In this sense, our focus on the institutional dimension should not be taken to mean that we underestimate the other dimensions influencing poverty, but as an intermediate position between a structural (more socioeconomic) and an individualistic view of the phenomena. Both elements play a fundamental role and the profiles of the socially assisted are the result of the complex process of social construction, which involves all dimensions. Socioeconomic restructuring and market-oriented reorganisation lead to greater vulnerability, while an individual lack of competencies is an obstacle in adequately facing the new context. Yet, their impact on living conditions is mediated by the institutional set-up, which distributes resources (or not) and prevents (or not) people beginning a path of downward social mobility. The institutional dimension acts as a filtering process, which transforms a larger or smaller part of the poor into assisted poor. In this sense, it is clear that in our analysis we are addressing only a subgroup of the poor, that is those covered formally by the measures and — among them — those taking up the available benefits. The process underlying the transformation of the poor into welfare claimants and beneficiaries can be approximately formalised in a chain of events on which specific measures (or their absence) have a greater or lesser influence, depending on the resources they can make available. Figure 4.2 shows a simplified version of this process, addressing typified events and listing examples of dimensions contributing to the pre-structuring of potential paths of downward (or upward) mobility. The perspective here is slightly different from that in Figure 4.1, as it attempts to give a more concrete insight into the process taking place. Specific events like family disruption, health problems/illness or unemployment may accumulate over time or may suddenly break down the existing balance between resources and needs. From the analysis of the ESOPO fieldwork material, in particular the interviews with beneficiaries, it emerged that events triggering downward mobility are related to three main spheres: (a) the labour market; (b) the family; (c) health. A fourth sphere, which may interact with the first three and which affects mainly migrants and refugees is that offorced mobility, due to wars, major national upheavals, political persecution or Social assistance dynamics in Europe serious economic poverty in the countries or regions of origin. Within these spheres, the degree of variation of events is relatively high, even though there are several elements common to the way in which they structure the specific condition of need that allow them to be grouped together. If we consider, for instance, the labour market sphere, a claimant might have become unemployed after a long working career iii the formal labour market, or else might have always been working in precarious jobs at the margins of legality, or have been outside the labour market altogether, as in the case of many lone mothers who apply for income support as a consequence of a marriage breakdown. Often events occurring in one sphere also impact on the equilibrium in other spheres. Serious illness can lead to job loss, family problems and a consequent divorce — or the other way around. According to the results of our fieldwork, the chain of events leading to the condition of need can be the result of three main processes: (a) it can be inherited from the family of origin, where it consolidated and has been transmitted from one generation to the next; (b) it can be the result of a major breakdown event, which tipped the existing balance between resources and needs; (c) it can be the result of a series of smaller breakdown events, which cumulate during a period of time bringing about a progressive but steady downward mobility. It is clear that the impact ol these events is manifold and depends on the resources the person can bring to bear. In Figure 4.2, several options are given as possible ways of coping with the 'crisis event'. These range from being embedded in rich and dense reciprocity networks to facilitated access to new job opportunities on the labour market, perhaps after a retraining course funded by a contribution-based scheme, as in many Scandinavian or Continental European countries. Social policies contribute to structure all these events; and even the perception individuals may have of the risks they are experiencing depends partly on the resources available. And individuals and households develop their coping strategies out of these perceptions. Figure 4.2 does not show all the possible options a person may have in managing conditions of need (the crisis) by accessing reciprocity networks and labour market opportunities directly. Again, this does not mean that we do not consider them relevant. On the contrary, they are often crucial. Rather, we consider them through their interaction with social assistance provisions, since the target of our analysis is people entering welfare. Furthermore, the profiles of welfare recipients depend, to a greater degree than might be expected, upon the institutional design of the benefits they can access, and the role played by the other dimensions is filtered through it. A first set of crucial dimensions of this design relates to the Social assistance dynamics in Europe conditions of access (certainty, budgetary constraints, subsidiarity, Stigmatisation, and so on), which determine — formally and in practice — the implementation of claimants' entitlements. To these dimensions we must add the degree of discretion social workers have in defining both access and the kind of intervention. The latter is further influenced by a set of crucial dimensions: the adequacy of replacement rates, the duration of payments in relation to the persistence of the condition of need, the existence of accompanying measures, their effectiveness, and so on. Altogether these dimensions — described in detail in Chapter Three — contribute to structure the profile of those who enter social assistance and, as we shall see in Chapter Five, also their patterns of recipiency over' time and chances of exiting. Who is more likely to be on benefits in the ESOPO cities? It is not our intention to argue that the social and individual characteristics of social assistance recipients6 depend entirely on the institutional design of social policies. The social assistance archives -we have used do not always contain sufficient data to prove such a hypothesis. In addition, the socioeconomic context plays a major role by adding a strong element of variation to the existing diversity in institutional patterns. More modestly, and cautiously, from the vantage point of a comparative perspective, we wish to understand to what extent the over or under representation of some categories is caused by the institutional design of the welfare scheme, and to what extent by the socioeconomic context. Of course, concrete cases are never completely clear-cut, and it is obvious that both dimensions interact strongly. If the institutional design and pattern of implementation were the same everywhere, the different profiles found in our study could be attributed largely to the effects of the socioeconomic context. In our sample, however, we not only have different national systems, but cases where cities share a common, nationally defined, institutional pattern, and cases where cities, although they belong to the same country and overall welfare system, do not have similar institutional patterns with regard to income support policies. Thus, inter-city differences in social assistance recipients'profiles must be interpreted carefully: they may depend either on inter- or within-country variation in institutional design or implementation patterns, or local social and demographic features. Our study allows us partly to assess the relative weight of these different dimensions, and more precisely, I 1A Why some people are more likely to be on social assistance than others the crucial role of the interaction between the institutional framework of income support policies and the modes of their implementation. Table 4.1 shows the distribution of first time recipients in the cities examined by selected demographic characteristics at the beginning of the 1990s. Data come from local social assistance records, which in some cases have been specifically reconstructed for this study. Thus, they include only the cities for which such archives were available7. Only in the case of the French cities do the data not regard first time recipients, but a sample of the total recipient population. It should be pointed out that the very existence (or lack) of archives — and therefore some form of recording and effort to keep track of beneficiaries, their characteristics and needs — is a significant feature of a policy and its implementation. It points to an attitude and willingness to monitor and be monitored, to be examined from the point of view of implementation and so forth, as well as to evaluate efficacy over time. The Swedish cities There are important similarities between the Swedish cities in terms of the characteristics of minimum income recipients entering the measure for the first time, for which there are socioeconomic and institutional explanations. Both cities have a disproportionately high number of immigrants who entered Sweden mainly as refugees, have residence permits and are trying to enter the labour market. These people generally come from a background of civil or ethnic war in their home country. Even though the main crisis event is their forced migration, they often accumulate several conditions of disadvantage. Examples are the lack of knowledge of the language, weak networks in the host country, inappropriate skills considering the conditions of the local labour market. For instance, if an immigrant has worked for 10 years in the public administration of a former communist country, their knowledge and practices are quite different from those common in Sweden. This brings about problems more to do with social integration and stigma than economic integration, which, as in France and partly in Germany, is granted institutionally (see Chapter Three; and Eardly et al, 1996), with benefits for immigrants being the same as for nationals. In both cities young people are highly over represented, even though they are often quite highly qualified. This may be at least partly explained by the high individualisation of the Swedish welfare state and cultural approach, which does not expect young people to be supported by their parents while looking for a job. At the same time, the comparatively low le 4 accords S: Distribution of first sg to main demographi ime social assistance recipients and total population within ESOPO cities : characteristics (%) Milan3 Tu irina Barcelona" Vitoria" Bremen3 Halle' Total Assisted Total Assisted Total Assisted Total Assisted Total Assisted Total Assisted Age (18-64) 18-29 26.0 18.8 27.4 26.0 25.4 15.4 30.5 28.8 26.9 54.4 23.8 na 30-44 30.5 43.3 31.0 36.8 33.1 35.9 36.5 45.2 33.3 33.2 33.4 na 4S-64 43.S 37.9 41.6 37.2 41.5 48.7 33.0 26.0 39.8 12.4 42.8 na Gender (18-64) Male 49.3 64.3 49.9 58.8 48.5 46.3 49.5 44.2 50.7 64.3 52.2 48.9 Female 50.7 35.7 50.1 41.2 51.5 53.7 50.5 55.8 49.3 35.7 47.8s 51.1 Nationality Nationals 95.1 86.9 98.3 84.7 98.0 95.0 98.0 95.5 86.4 74.2 98.6 92.0 Non-nationals 4.9 13.1 1.7 15.3 2.0 5.0 2.0 4.5 13.6 25.8 1.4 8.0 Household size (i) 1 32.0 27.3 33.5 27.3 18.1 45.1 37.1 35.1 44.6 42.5 33.9 na 2 28.0 21.2 27.2 17.6 27.8 18.8 28.6 19.1 32.6 19.9 31.2 na 3 21.2 21.4 19.8 14.5 22.0 14.4 15.8 20.4 11.7 18.7 19.7 na 4 or more 18.8 30.1 19.6 15.4 32.1 21.7 18.5 25.4 II.1 18.9 15.2 na Marital status (i) Unmarried 39.8 31.0 38.0 29.8 41.8 43.8 47.3 35.9 30.2 36.9 38.0 na Married 47.5 32.6 49.5 26.3 47.9 23.8 46.6 27.4 51.4 40.9 47.0 na Separated 1.7 21.4 2.5 10.2 1.6 23.8 na 29.4 (2) 12.1 (2) na Divorced 1.8 7.6 1.6 4.4 0.9 4.8 na 3.1 7.7 7.9 m na Widowed 9.2 7.4 8.4 5.4 7.8 3.9 na 4.2 10.7 2.2 15.0 na Target groups Single adults 32.0 27.3 32.3 23.3 18.1 45.3 37.0 35.1 42.8 41.2 33.9 na Non-nationals 4.9 13.1 1.7 15.3 2.0 5.5 2.0 4.4 13.6 25.8 1.4 na Lone mothers 7.7 15.4 6.2 8.5 7.3 26.2 7.0 19.7 4.33 10.6 na 24.6 Couples with children 34.4 26.3 36.2 16.6 45.0 14.5 69.0 19.5 14.5 15.7 na 13.6 Number of cases 1,31 1,979 791 979,839 1,230 1,643,542 585 206,1 16 385 549,357 849 303,000 652 continued Table 4.1: Distribution of first time social assistance recipients and total population within ESOPO cities according to main demographic characteristics (%) (contd) Gothenburg" Helsingborg" Lisbon" Rennesb Total Assisted Total Assistec Total Assisted Age (18-64) 18-29 30-44 45-64 Gender (18-64) Male Female Nationality Nationals Non-nationals Household size 1 2 3 4 or more Marital status Unmarried Married Separated Divorced Widowed Target groups Single adults Non-nationals Lone mothers Couples with children 31.3 35.2 33.5 50.3 49.7 89.0 11.0 18.2 26.6 20.6 34.6 43.6 43.2 n.a. 11.7 1.5 18.2 I 1.0 51.1 35.1 13.8 53.2 47.7 66.7 33.3 67.06 15.9 8.5 8.6 19.0 67.0 33.3 13.0 12.0 28.1 34.1 37.8 49.4 50.6 93.2 6.8 13.8 27.2 21.8 37.2 36.3 49.8 na 12.0 1.9 13.8 6.8 59.0 27.1 13.9 53.3 46.7 78.9 21.1 61.56 16.4 13.5 8.6 59.8 20.9 17.2 16.0 23.4 30.5 46.1 46.6 53.4 23.9 28.4 21.3 26.4 28.1 53.3 2.6 4.3 11.7 23.9 1.8 1.9 50.4 15.7 36.4 48.0 32.8 67.2 95.4 4.6 23.3 20.2 24.7 31.8 27.3 38.9 13.6 11.6 8.6 23.2 4.5 23.7 31.8 Saint Etienneb Total Assisted Total Assisted 32.3 4I.88 33.1 39.1 34.6 19.1 48.5 51.5 89.1 10.9 34.6 29.7 15.3 20.4 33.2 52.5 (7) 9.7 4.6 34.7 10.9 6.7 31.0 48.0 52.0 76.3 23.7 44.9 18.8 12.7 23.6 57.1 27.2 (7) 11.8 3.9 44.9 23.7 40.9 30.1 29.0 95.9 4.1 37.6 29.3 15.0 18.1 45.1 43.5 (7) 4.7 6.5 37.6 4.1 7.8 26.8 43.08 42.1 14.9 45.7 54.3 91.7 8.3 47.5 24.2 14.7 13.6 66.1 13.6 (7) 16.9 3.4 47.4 8.3 Sources of data Total population: If not specified the source is the relevant census of 1990 or 1991. Germany: Mikrozensus, 1991. Portugal: Lisbon population, 1992. Assisted: (a) ESOPO longitudinal sample on first time recipients. (b)Oberti(l998) (c) 10% sample 1991 cohort, OlkandRentzsch, 1997. Notes (1) Differences to 100% missing data. (2) Data for separated and divorced included in widowed. (3) Households on household total. (4) Year 1994. (5) 20-64 age bracket. (6) In Sweden households are defined differently (see text). (7) Divorced and separated are considered together. (8) 21-29 age bracket. Data on target groups do not equal 100% because they do not include all beneficiaries and in some cases overlap.Year of reference when not specified differently is 1991. Number of cases 449,200 2,213 14,000 244 663,394 199,464 267 197,494 122 Social assistance dynamics in Europe satisfactorily calculate the number of people entering the measure more than once: entrants in a given year do not all represent first time entrants. A number may be returning after an interruption. Thus the cohorts of beneficiaries in the CNAF sample are not fully comparable to those in our study — they may over represent difficult cases, that is those without enough resources to remain off social assistance. It is likely that data without these censoring problems would indicate an even higher similarity to those of the north European cities. ''The medians of the first dependence period read: Barcelona 34.4 months, Bremen 8.1 months, Gothenburg 10 months,Helsingborg 6.4 months,Lisbon 33.5 months, Milan 6.5 months,Turin 10.7 months andVitoria 15.4 months. 10 In Barcelona the amount of income support is lower than inVitoria. 1' It should be pointed out, however, that at least in Italy separated and divorced (not widowed) lone mothers have on average higher labour market participation and employment rates than married ones. 12 The median for the cash episode among single parents is significantly higher than for other recipients in Helsingborg, while the corresponding is not found when analysing dependence periods. The reverse (statistically significant longer dependence episode, but not cash episode) applies to Vitoria. 13 This rule may also be applied to EU citizens. '"' There are very few foreigners in the samples for Lisbon andVitoria. i5The greatest advantage of this model, in comparison with the survivor function, is the possibility to consider many different co-variables and estimate their effects on the rate. Besides, we can see the significance of the effects on the rate. The significance depends on the number of events (transitions out of social assistance) and the distance in the calculated rate. Therefore, the cities whose data shows a higher number of cases would structurally reach a higher significance. To avoid this problem, we have weighed the cases so as to get a total of 500 cases for every city. "'These results on significant coefficients for variables measuring type of household are confirmed when models explaining the first dependence period are estimated. 17 The estimate for Lisbon is based on a small number of observations. IK However, when analysing dependence periods the ranking is slightly different, that is: Barcelona, Lisbon,Vitoria, Turin, Gothenburg, Bremen, Helsingborg and Milan. 234 SIX Chiara Saraceno Social assistance In the face of changing demands for Long-term unemployment among adults, delays and constraints in entering the labour market among the young, impoverishment of women and children due to the gender division of labour, as well as growing marital instability and the weakening of father-child ties', are all phenomena that challenge the traditional solutions offered by social assistance. In countries where income support measures in the form of a basic or minimum living allowance have been developed as a temporary stopgap for those who have lost the social security protection linked to their status as workers, changes in the labour market conditions and in life course patterns are transrorming both the type of beneficiary and their experience. Together with the growing number of unemployed people who now have less opportunity of finding a new job due to age or lack of required skills, and the growing number of young people who have difficulty even entering the social security protected labour market, there is also a growing number of people who find themselves with little or no income or social security protection for reasons not directly linked to the labour market. Among these are unmarried, separated or divorced women with small children who invested in raising a family rather than in paid work, young people who have adopted self-destructive behaviour, families who are over-strained by the long illness of one of their members, and immigrants who encounter obstacles in their efforts at social integration and social mobility. Income support may thus be received by populations who are dis-homogeneous not only with regard to their life course, needs and personal resources, but also with regard to the perception of their situation, as well 235 Social assistance dynamics in Europe as the strategies developed2. There may be individuals and groups who are ashamed to resort to social assistance and who ask for help only when they have exhausted or severely weakened their own and their social network's resources. In contrast, other individuals and groups develop great skill and ingenuity in trying to get hold of some kind oi entitlement to social assistance. Together with those who mix dependency on social assistance with more or less formal work activities and income, and those who use social assistance as a temporary measure while they look for a job, there are those who use social assistance as a substitute for searching for a low paid job. The proportion in which these different groups may be found in social assistance schemes, however, does not depend solely on their incidence in the population in a given country or city. It depends also on the working of the social assistance schemes themselves: their filtering mechanisms, time rules and so forth. Thus, social and political awareness of the heterogeneity of the potential claimants for some kind of support may differ cross-country and cross-city, depending on the country and city-specific institutional definitions of who is entitled to support. For the same reason — the impact of institutional definitions — changes both in labour market opportunities and in social security regulations may increase this heterogeneity of social assistance recipients, thus also of trajectories within it. Individuals and families who in the past would have received non-means-tested unemployment benefit or a disability pension, may now have to rely on social assistance for shorter periods, due to changes in the criteria governing the entitlement to social security provisions. Furthermore, the growing number of non-standard labour contracts that, particularly in Southern Europe, are mainly concentrated among adult women and the young of both genders, is creating a subgroup of workers who by definition are less protected by social security measures. This is a particular risk in those countries in Continental and Southern Europe where social security is highly categorical and linked to the workers' status and seniority. Not being entitled to unemployment indemnity, these people may have to make recourse to means-tested income support as a last resort. But in some countries, particularly in the Southern European ones, they may not even be entitled to that, unless they are very poor, isolated, or suffer from some severe additional personal or social disability. Further, in the redrawing of boundaries between social security and social assistance, two, somewhat contradictory, processes seem to be going on. On the one hand, there is some kind of reshuffling or redistribution between categories of welfare state recipients, particularly between social Deconstructing the myth of welfare dependence security and social assistance. On the other hand, there is a refocusing of social assistance towards measures that are explicitly work-insertion oriented. The former process, due to the pressure on social security budgets caused by high rates of unemployment, tends to reduce social security coverage, particularly in the case of long-term unemployment (and also invalidity), moving people more quickly to means-tested measures. Moreover, the boundaries between social assistance and social security are becoming blurred. The most explicit move in this direction occurred in the UK when the Jobseeker's Allowance replaced both Unemployment Benefit and Income Support for working age applicants. The result was a blurring of the boundaries between the two groups of beneficiaries, submitting both to the same kind of close monitoring. Also, the requirement to undertake some kind of work in exchange for benefits — or work/are — is increasingly being attached not only to social assistance, but also to social security benefits in many countries. This goes well beyond the traditional expectation that unemployed people, especially able-bodied social assistance beneficiaries, accept reasonable job offers3. We will return to this in the section on 'Explicit and implicit assumptions in social integration discourses'. The potential and constraints of comparing social assistance policies The growing heterogeneity of those requiring some kind of income support, as well as their number, is occurring in a policy context marked by an increasing demand for the evaluation of social policies. Two distinct objectives spur this demand: the control of public social expenditure and the improvement of efficacy. The evaluation of income support policies shares this concern for accountability. Comparative cross-country evaluation is perceived as a useful tool in the sense of'learning from other countries' and looking at 'best practices'. Yet, it encounters specific and serious difficulties that should be clearly spelled out. This is not to deny the possibility of comparative evaluation, but to clarify the boundaries and limits of such an endeavour, as well as the kind of data needed. A first difficulty concerns the objectives themselves, both at the institutional level (as defined in laws and regulations) and the implementation level (as defined and perceived by social workers and potential beneficiaries). Bouget and Nogues (1993) claim that the sheer multiplicity of goals in national social policies in the field of income support makes it virtually impossible to identify any common trends, priorities or even philosophies. This is possibly an exaggeration. Yet, as Social assistance dynamics in Europe we have seen throughout this book, countries differ considerably in their definition of the 'deserving' and the 'undeserving' poor, their definition of need and the risks of social exclusion, as well as their definition of what is necessary to achieve social integration, and even the-meaning of social integration. We have seen that in some countries (for example, Portugal, Italy, and to some extent Spain) only those who are perceived as being at high risk of social exclusion, in that they are extremely poor and/or thought to be unable to take care of themselves, are entitled to income support. In other countries (Sweden and Germany), the mere lack of income entitles one to income support, without any judgement being made on the risk of social exclusion. A second difficulty concerns the plurality of institutional actors involved in tMs kind of policy: from the State to regional and municipal institutions, administrators and social workers, as well as public, non-profit and sometimes also market institutions. Each may have a different perception or definition not only of the goal of the social assistance measure in which they are involved, but also of the beneficiaries. And they may combine and interact differently cross-country as well as cross-city, producing distinct patterns of implementation and different definitions of poverty and social exclusion. Evaluation, and the evaluators themselves, including the European Union4, are becoming institutional actors in so far as their requests (for data and information) and the results of their exercises influence the institutional setting, the self-perception of the actors and, more generally, the perception of the policy in question by the public. From this point of view, it is correct to say that evaluation is never a neutral process, however objective it may try to be (see also Bouget and Nogues, 1993; Rossi and Freeman, 1993). A third difficulty concerns the various 'policy packages' (of which income support policies are usually part) being developed to combat, directly or indirectly, poverty and social exclusion. The combination of specific access rules, coverage, duration and generosity of unemployment indemnity lead to considerable differences in the timing of entry into social assistance, as well as in the characteristics of its beneficiaries. The generosity of child allowances in one country may keep families with children off social assistance. In another, the existence of specific assistance for lone mothers may help remove the need for this category to resort to general social assistance for some period. The rules in other parts of the policy package therefore have the effect of selecting demographic or other features (for example, age of child) of beneficiaries differently from one country to another, even before the specific filtering mechanisms of the measures analysed and compared operate their own selection. 238 Deconstructing the myth of welfare dependence As a result of these three difficulties, a fourth arises. A cross-country and sometimes even cross-city differentiation in the definition of eligibility for social assistance resulting from the variety of goals, policies and patterns of implementation, produces fundamental differences in the population of beneficiaries. This in turn has consequences for the performance of the social assistance measures themselves. Our research clearly shows that legal and/or de facto eligibility filters have a far more important impact than any other dimension. They affect not only the demographic and social characteristics of beneficiaries, but also the average duration of welfare dependence, the risk of recurrence, the chance of successfully exiting, and so forth. -■ To these methodological and conceptual difficulties we should add another, more technical one. Cross-country and cross-city comparisons and evaluations can be performed only if there exist good longitudinal local data, collected and recorded according to standard methods. This is an almost trivial observation and is of course valid for any evaluation or comparison, but it has specific importance in the case of social assistance and should not be underestimated. Within a social assistance system, the collection of data on beneficiaries is part of the process of policy implementation itself. Social workers and administrations routinely record a variety of biographical data on beneficiaries. These, however, are not always suitable for monitoring, reviewing, evaluating or comparing purposes. This may be for three distinct reasons. First, the data are rarely standardised. Second, even when they are, they respond more to the needs of accounting (how many interventions and for what cost), than the needs of monitoring or evaluating the impact and outcome of the policy (how many individuals or households are affected, for how long and to what effect on their living conditions). Within-city and within-country monitoring and evaluation, as well as cross-city and cross-country comparison and evaluation, rely on having access to standardised long-term data, which allow the possibility of following the history of individual social assistance recipients. They also require accurate data on policy implementation. The lack of the former sort of data is why we were unable to compare social assistance 'careers' in all the cities involved in this study. Not all cities possessed such information, even in an elementary form, while in other (French) cities they existed, but were unavailable to outside researchers. As for the data on patterns of policy implementation, they must be carefully constructed through a variety of means and sources. The third reason is of a different kind, as it involves an ethical issue. The need to keep individual data over time and to follow-up beneficiaries after they have left social assistance, in order to monitor and evaluate the Social assistance dynamics in Europe performance of social assistance, is to some degree in contrast with the right to privacy. And this is a right that should be particularly protected in the case of those who are already highly vulnerable socially. This is precisely the reason why social assistance records data have not been available for the French cities. While the solution to the first two problems, though often complicated, is basically a matter of administrative culture and choices, the third problem is one that is likely to become more serious in the future. Awareness of the importance of monitoring and evaluation for social policy making and awareness of the inviolability of individual rights — including that of privacy - are both growing, but pulling in two different directions. Possibly the EU is the level at which this apparently intractable contradiction should be addressed, not so much for the sake of research as for well informed and self-critical social policy making. The recognition of these constraints and difficulties has enriched the framework of our evaluation and also our methodological approach. We have not simply considered and compared the performance oi different income support measures in terms of statistical output (how many people, Avith what characteristics, and for how long). We have tried to reconstruct and compare the overall mechanisms, as well as the sets oi actors involved, looking at the whole context within which and through which income measures are constructed and implemented. We have therefore used a local, multiple and integrated perspective, also considering how social assistance is experienced by beneficiaries. The countries have not been compared on the basis of aggregate data, but in terms of the local systems, the workings of which we have, at least partially, attempted to reconstruct. In the following sections, we briefly summarise our main findings, which we feel are relevant both from a policy and a methodological point of view. Locaf contexts, welfare mixes and the diversity of the third sector The ESOPO research findings allow us to appreciate to what degree it is important when analysing poverty and policies to combat poverty and social exclusion, to not only grasp the relationships between family, State, market and third sector but, even more, to understand and distinguish the forms and dynamics of each of these four dimensions as well as their interaction. In the first place, our study has confirmed the crucial role of the family in the national and local welfare mixes. Legal and de facto expectations Deconstructing the myth of welfare dependence concerning family and kin solidarity, that is the degree of familisation or vice versa de-familisation (Esping-Andersen, 1999; Gallie and Paugam, 2000; Saraceno, 1997), of rights and of access to resources vary greatly across our cities, contributing to the diversification both of the structure of the support package and of the selection mechanisms. At the same time, within the same formal social assistance system, the presence of supportive family and kin may make a huge difference for the well being of an individual and household. In the second place, our research shows that the usual distinction between State, market and third sector is over simplistic. First, the State itself must be considered in its various forms, as well as its different levels. These have varying importance and degrees of autonomy in different countries. Public institutions — whether State, regional or municipal -have a hfe and logic of their own, with specific professional as well as institutional interests and cultures. Coordination of different institutional levels and actors of the public administration may be as difficult as that between public and private ones. From this point of view, the different pattern of collaboration between State and municipal institutions in Saint Etienne and Rennes is a particularly telling example. Of course, the market too has its plurality of actors, which may vary considerably from one place to another. This is particularly so for the unofficial, informal market economy, as indicated by the comparison between the Portuguese cities and the Swedish ones, as well as between Barcelona andVitoria in Spain. But, second, possibly one of the most interesting findings of our study at this level concerns the crucial and diversified role of the third sector. The third sector is usually understood as being comprised of such diverse actors and agencies as churches, voluntary associations, non-profit organisations, and so forth. And equally simplistically, it is often assumed to be, together with family and kin, the privileged arena of social integration. Yet we have found that both the specific third sector actors present in a given context and the way that they are mobilised may have a very different impact on social integration. Even less does the third sector automatically make up for the lack of a well-developed public one. Thus in Cosenza, the city in the ESOPO sample where the public sector is least developed, the presence and role of the third sector is negligible and overall integration in and of the local community is weak. Although it may happen that a strong public sector monopolises all the collective resources for solidarity existing within given national and local cultures, it seems more difficult for the reverse to occur, at least in our sample Social assistance dynamics in Europe cities: where the public sector is weak, the third sector develops and organises with difficulty. Moreover, there are great differences not only in the role and resources of third sector actors, but also in their social status. We find differences in the human and financial resources they dispose of, as well as in their social or political standing. Highly organised institutions such as Caritas in Italy (or La Misericordia in Lisbon) have very little in common with small groups of dedicated volunteers who, relying on their own resources, go out at night in search of the homeless. The differentiation within the third sector has been increasing in recent years (Gidron et al, 1992). This is partly because new actors have entered the fight against poverty and social exclusion, often catering for particular categories (for instance, Third World immigrants, children, drug addicts, HIV bearers and AIDS sufferers), partly because public institutions have called on them as the most adequate partners for dealing with poverty and social exclusion, particularly with regard to social insertion programmes. In all cities we found a multiplicity of intermediary structures which are crucial in anti-poverty policies, sometimes in formal collaboration with public structures, sometimes operating in relative autonomy, and sometimes in competition. Though not part of the public system, they are often formally integrated into it. Further, they are being increasingly acknowledged as full partners in policy making, both at the national and at the EU level. Actually the EU has given a strong impulse to the institutionalisation of social partnership - including not only trades and enterprise unions, but also NGOs — not only at the EU, but also at the national level, requiring that they be involved in the open coordination process. Yet, as we have pointed out in Chapter Two, this increasing role of third sector institutions and actors is developing within nation and even local specific traditions, range and type of actors and so forth. This in turn makes for quite different kinds of welfare mixes and more generally of national and local patterns of governance. It makes also for different patterns and understanding of social citizenship, at least from the point of view of beneficiaries. The greater attention for the relational dimension of support, together with the higher flexibility and diversification of provisions, which are usually associated with the action of third sector agencies compared to public ones, are often premised on specific values that identify a given group and associations. This is particularly, but by no means exclusively, the case for religious (as well as for ethnic) associations. These values and shared identity constitute a crucial motivational background for these associations and their social workers; they may also be an integrating resource for beneficiaries. But they may Deconstruďmg the myth of welfare dependence also be perceived by the latter as an additional burden: an obligation or price they have to pay in order to be supported. Further, particularly in places where the public sector does not play an important role in defining criteria for entitlement and provision, the distinction between charity and citizenship rights may be further blurred by the intermediating role of particularistic value laden agencies and actors. In our sample, the Swedish and French cities are those with by far the most regulation by the public system and the weakest involvement of intermediate structures - both family-kin and third sector agencies - in providing social assistance. This is, of course, premised on a greater individualisation of entitlements. But it may also produce, or ignore as a specific need and lack of personal resources, social isolation when an individual or household for some reason is cut off from family and community ties. At the same level of poverty, this may lead to greater and more rapid isolation and alienation of those who may not count on an informal network of relationships and support, even if they do receive income support. Despite reunification, the two halves of Germany still display considerable differences in their conception and organisation of policies for dealing with poverty, and there are important local variations even within what used to be the German Federal Republic. Bremen corresponds to a model in which a wide network of local intermediate organisations - trades unions, professional or religious associations - is strongly coordinated to public welfare. Halle still shows signs of the old socialist system, which tended to marginalise volunteer, self-help, nonprofit and religious associations. Even if the web of charitable organisations in the city is extending and intervention growing, it has not yet reached the density and diversity of the network in Bremen (where more public resources are also available). In Italy and Spain, the greater autonomy of the third sector is related to a process of institutionalisation, where great efforts are made to involve non-profit institutions and even to support their development. In Turin andVitoria, both the secular and religious third sector agencies are being increasingly integrated into the institutional arena of social policies. This pattern is present also in Milan, where, however, collaboration between the public and third sector seems to occur more through an implicit and explicit delegation of responsibilities by the former to the latter, than through coordinated planning. The Italian case demonstrates not only the different patterns of collaboration between the public and private sector that may develop on the basis of the features and history of both Social assistance dynamics in Europe sectors. It also demonstrates how interdependent these two, apparently separate, histories are. As for Portugal, it is undergoing a process of profound modernisation that should lead the country to break with the traditional model where assistance for the poor was closely linked to the notion of charity and private aid. There are, however, different paths to institutionalisation even here. In Lisbon, social assistance policies are run by a non-profit body of charitable origin that is legally controlled by the State, but organisationally and financially very independent. In Porto, policies are run by a public regional social services centre, with no cooperation with other public or non-profit bodies. Finally, it should be pointed out that the third sector may play a twofold role in combating, or preventing, poverty and social exclusion. On the one hand, it may substitute or integrate public support policies, either through the provision of resources or through the provision of services. On the other hand, it can provide jobs, and thus a way out of economic poverty, particularly for those who suffer from temporary or permanent vulnerabilities in the main labour market5. This latter role is increasingly being focused on both in European Union policy documents and by national and local regulations. Social assistance careers as the outcome of biographical and institutional constraints Our study has demonstrated that both the demographic and social profiles of income support beneficiaries and the form of their 'social assistance career' is highly dependent on the way conditions for entitlement are institutionally defined and practically implemented. As a consequence, cross-country and cross-city variation cannot be easily and unambiguously interpreted in terms of the greater or lesser efficacy of the welfare assistance provided. Conversely, similar outcomes do not automatically have the same meaning. This apparently negative finding (with regard to the possibility of performing a comparative evaluation) has prompted us to look deeper at the differences and their consequences, allowing us to understand better the workings of a given measure in a given context, as well as to dissolve long-held stereotypes. Our data offer sufficient grounds to disprove two of the most common critiques addressed to income support measures. Firstly, that they create long-term dependency and secondly, that the more universal and generous the support, the more likely it is that people remain on social assistance for a long time, becoming dependent on it. We found, in fact, that only Deconstructing the myth of welfare dependence a minority of beneficiaries in all cities remain for a long period on social assistance, irrespective of its characteristics and the rules for entitlement. Of course, this is not a univocal indicator of success. It might even be suggested that longer periods on social assistance could, in some cases, be more efficacious than forced short periods, in so far as they would allow the investment of time and energy in building up personal resources and assessing possible opportunities. Yet our findings show that the widespread idea that most social assistance recipients are long-term ones simply does not correspond to reality. Previous studies that were geographically and, to some extent, also methodologically more limited than ours (for example, Duncan, 1984; Paugam, 1993; Walker and Shaw, 1998; Leisering and Leibfried, 1999) have already indicated that there is no evidence that social assistance per se creates dependence. Our study confirms this finding on the basis of a much larger comparative sample, while at the same time qualifying it. On the one hand, it supports the thesis that long-term dependence, when it occurs, is less a consequence of the'corrupting'impact of social assistance — particularly its generosity or lack of rigid constraints — than of the 'original'features of the beneficiaries. Those who remain for long periods on social assistance, or have recurrent relapses, are likely to have some specific disadvantage, such as being of a mature age, in bad health, female, with heavy family demands, low education and skills, and so forth. Thus, the more common these features in the. population receiving income support, the more likely are long careers in social assistance, when these are not deterred by a time limit imposed by regulations. The concentration of specific vulnerabilities and weaknesses in the population receiving social assistance is, in turn, a result of the interplay between the social, economic and demographic features of the national and local context, as well as the rules governing income support schemes, particularly those pertaining to generosity and universalism. There is a higher concentration where national and/or local economies are weaker and social assistance supports only the very poor, or those who, in addition to being poor, have some specific kind of individual or social vulnerability. This combination holds for all our Southern European cities, where social assistance beneficiaries tend to suffer not only from income poverty, but also a more general situation of social and biographical deprivation. Only in Milan, and to a lesser degree Turin, does this combination not give rise to long social assistance careers, either continuous or with interruptions. But, as we have indicated, this is no proof of the efficiency and efficacy of the system of support in these cities, rather of the rules and constraints applied. With the exception of given categories (for example, lone mothers, 8246 Social assistance dynamics in Europe people with special needs), income support in these two cities, and particularly in Milan, is interrupted after an allotted time has expired; beneficiaries may not return within a predetermined time limit, irrespective of need. Thus, they often turn to the voluntary and charity sector. The highest dependence rate was found in Barcelona and Lisbon, where the combination of close targeting of beneficiaries (which restricts entitlement to the extremely poor and vulnerable), the small amount of benefit, and the absence of a time rule, led to longer than average periods of social assistance. In this case, although beneficiaries include a large percentage of people with great difficulty in becoming self-supporting, the benefit is too low to give them a chance to make a fresh start. They are hence compelled to find ways of integrating this benefit with other resources, often in the informal economy, where it is easier for people with low skills or personal difficulty to find an odd job. However, this does not offer them the chance to become self-supporting. From this point of view, the expression'social assistance dependence'is only partially correct, in so far as social assistance alone does not give these individuals or households sufficient means to survive. One of the main findings of our study is that close targeting and low benefits create a population of beneficiaries characterised by a high degree of vulnerability and difficulty in becoming completely autonomous from social assistance. This kind of population is of course present everywhere and needs special attention. But when it comprises the majority of social assistance beneficiaries due to strict selection rules, it becomes part of the definition of social assistance itself. The lack of efficacy of "welfare measures in rendering this kind of beneficiary autonomous sets up a vicious circle, reinforcing a negative pessimistic view of both policies and beneficiaries. It also de-motivates social workers, increasing their own vulnerability to professional burn out. It therefore constitutes in itself a path to marginality and social exclusion. This may cause different reactions, according to the local political culture and economic resources. Thus, indefinite (partial) dependence may be accepted by some social workers and administrators without prompting any effort to redefine the approach. Alternatively, a revolving door or shifting mechanism (from public to private social assistance) may develop, again without prompting policy innovation, even with regard to this particularly vulnerable population. In contrast, where benefits are relatively generous, beneficiaries more 'mixed' and the chances of success higher, it is less likely that the presence of a certain proportion of 'difficult' beneficiaries will have a strong stigmatising effect on the whole population of recipients, and on the institution of social assistance itself. "tAi. Deconstructing the myth of welfare dependence Universalism and generosity would therefore appear not only more appropriate to a citizenship culture, since they lead to better social integration, but also more effective in the medium-long term: in so far as they prevent people from exhausting their resources and starting social assistance too late to be successfully supported. This finding, however, needs to be qualified. It is certainly true that within these systems we find a substantial quota of so-called 'bridgers', that is people who use income support as a 'bridge' between periods in which they are working: in other words to fill the gap when they are temporarily out of work and not covered by social security. Social assistance acts for them as a temporary cushion while they look for a new job. However, among these, some people may be subject to some kind of'revolving door' mechanism, albeit in a highly structured and protected way. Given the interplay between regulations concerning unemployment indemnity, socially useful jobs and income support, in Sweden and Germany, but also Vitoria, one may go through (or be pushed through) the whole system — from work to unemployment indemnity to social assistance — and then start again". It is worth noticing, however, that within this system, and particularly in Sweden, social assistance does not keep beneficiaries off the labour market. Rather it tries to maintain continuity between being in work and being out of work, as well as between being on social assistance and on social security. This said, both in the less generous and the more generous systems, the presence of a group of beneficiaries with a higher risk of long-term dependence on social assistance cannot easily be dismissed, and is not susceptible to a single solution. It points to a series of different problems that have differing weight in the various national and local contexts: labour market restructuring and high unemployment rates, lack of skills, low paid jobs, the inadequacy of support for families with children, the gender division of labour within the family and its consequences for women's labour market participation. No single policy can of course address all of these. To some degree, measures for dealing with the poor are really trying to cope with failures in other systems: the labour market, social security, the family and school. Thus, it is empirically unfounded to base any request for social assistance restructuring on the assumption that it produces dependence and that most beneficiaries would remain on social assistance if not adequately monitored and pushed out; yet social assistance systems must also deal with the presence of a group who either needs it for a long time, or who cannot be moved out of it without requiring both additional investment in accompanying measures and some change in other areas. 247 Social assistance dynamics in Europe Explicit and implicit assumptions in social integration discourses The last observation points both to the crucial role of differentiated, accompanying measures in facilitating the social integration of the poor and to the implicit, and sometimes conflicting, assumptions concerning what social integration, or social inclusion, is about, at least with regard to socially vulnerable groups. We could say that while the countries of Southern Europe tend to emphasise the role of families and communities in providing social integration, even in conditions of long-term unemployment and poverty, the more universalistic systems share an idea of inclusive citizenship mainly based on participation in paid work, for both genders. In the Swedish cities, social integration is in fact easily translated into 'having a job and becoming financially self-sufficient'. Although the UK was not part of our study, the most recent developments in the British welfare system make this idea of social inclusion through inclusion into the labour market most explicit. Yet, this form of social inclusion is only apparently simple to achieve. First, it suggests that jobs - and more so adequately paid ones - are available, and that the problem lies entirely with beneficiaries: either their lack of willingness to work or their lack of skills. Second, it tends to underestimate the amount and value of the unpaid work that is mostly performed by women. Finally, it presupposes that work cultures and willingness to work for wages (and Stigmatisation of social assistance) are the same across social and ethnic groups, and are, to some degree, 'found in nature'. Many experiences testify to the contrary. The 'work ethic' that is part of the tradition of the industrialised world has been constructed through a long and often hotly contested process, in which different work cultures and different ideas of social obligation have competed. We can find traces of this in the unresolved conflict between the needs and ethic of care, and the ethic of paid work, which underlies gender conflicts. It also lies behind the increasingly fragile balance in the family's division of labour and the concept of obligations of family and kin. The very different way in which lone mothers with small children are regarded in Sweden and Germany or France for income support purposes, reflects different views of the social obligations of this group of beneficiaries, involving the balance between caring and working for pay. The UK itself, in its move towards redefining lone mothers as breadwinners rather than carers, is using the carrot of improved benefits (including money to pay for childcare), rather than the stick of enforced obligation to be available for work, indicating that social inclusion involves 248 Deconstructing the myth of welfare dependence something more than simply being moved into paid work (for this group at least). Thus, in order to give people an incentive to work, it is not always sufficient to offer them jobs (assuming there are jobs to be offered). These must be perceived by the beneficiaries as providing a good opportunity for their life and family. This in turn presupposes not only a specific cultural outlook and perspective, but also three other things: that the person offered the job has enough training to keep it, that caring obligations and needs are adequately taken care of, and that the job pays enough and offers enough security to be worth taking. The last aspect is particularly relevant in contexts, such as Southern Europe, where the offer of poorly paid and temporary jobs in the official labour market may be in competition with extended family solidarity, and also work in the informal economy. Since time spent in unemployment increases entitlement to social assistance benefits and to being hired in the public sector, taking a poorly paid and insecure job may appear a move that is too risky and costly7. In this case, it may seem more rational to remain in the informal economy, with or without social assistance. As for the idea that any job is better than no job, our data offer some support to Atkinson's (1998b) and Paugam's (1997a) argument that this is a very simplistic view, and that some jobs might lead to greater social and professional exclusion than social assistance itself. Forcing a person to take any job may have negative effects upon their skills, and hence the ability to stay in the labour market. This is particularly true when the unskilled job is not protected by social security. Of course, this general argument must be qualified with regard to specific individual circumstances. Yet it appears that the risk of the revolving door mechanism is greater when the 'any job' rule is strictly enforced. It can make a great difference when, as in the Swedish and German cities, even temporary jobs offered to social assistance beneficiaries give the same entitlements to social security — in terms of pension contributions and unemployment indemnity — as 'normal'jobs. By offering an individual the real status of worker, he or she is at least temporarily reinserted in the circuit of social security. Paugam (1997a) defines this as a form of disqualified integration' (if the job is unskilled), in contrast to the 'compromised integration' offered by unskilled jobs in the informal economy where there is no social security coverage. It should be added that integration achieved by accepting any job (with social security) may work better in places and within social groups who share a strong work etliic. From this point of view, it is interesting that notwithstanding the generosity of the system, Swedish beneficiaries Social assistance dynamics in Europe tend to spend less time on social assistance precisely because they have the view that it is. one's responsibility, as well as one's right, to work. In Sweden, unlike many other countries, this attitude is shared by lone mothers of small children, who are not covered by special provisions nor temporarily exempted from the requirement to work. The combination of widespread childcare services and a widely shared gender model in which motherhood and paid work are not in contradiction, renders this group less vulnerable than in other countries (where lone mothers tend to become poor because of lone motherhood, not because of loss of a job); but it also puts pressure on women to exit welfare quickly by finding paid work, irrespective of their mothering obligations and desires. Interestingly enough, the same attitude can be found among lone mothers in Halle, who grew up with the same gender model as their Swedish counterparts, but now find it in contradiction to prevalent models in the reunified Germany where they are incorporated into a system with a very different set of implicit and explicit expectations about gender, social assistance and the labour market. It is much more difficult for lone mothers in Halle to exit welfare than in Sweden, although they share the same expectations and values. At the same time, it is likely to be more difficult for lone mothers in Halle than in Bremen to accept their experience of receiving social assistance as justified by their caring obligations, rather than by their joblessness. Thus, the interplay between local and individual cultures concerning work and family obligations and local opportunities must be taken into account when defining options and projects for social assistance beneficiaries. It is important to help them orient their expectations and to develop strategies so as to avoid social assistance clashing with their personal and social identity, further contributing to a feeling of social exclusion. The concept of a 'contract' as opposed to a one-sided obligation has been developed precisely to avoid these rigidities and risks. It has been introduced as a policy instrument, particularly in France, to express the mutual commitment of the community and the beneficiary towards achieving social integration and an improvement in the beneficiary's situation. In the British system, it has been phrased in a slightly different way, as a 'new deal'. But in both cases, the responsibility of the parties — the beneficiaries and the State or community — are stressed. Yet, even this idea of striking some sort of contract, making some kind of deal with those receiving social assistance is not without problems. First, it implies that there really are alternatives or options, which again is not always the case, due not only to the history of the individual, but also to socially structured constraints. Second, it implies a symmetry 250 Deconstructing the myth of welfare dependence between the contracting parties. This presumes, on the one hand, a capacity for care and attention in social workers, as well as cooperation between the various local actors in developing specific opportunities. On the other hand, it presumes that the beneficiaries themselves have sufficient human capital to develop projects and to negotiate with social workers. In its actual implementation, this policy can easily be transformed either into a routine procedure, or a form of social control on recipients, or even a new instrument of discretionary treatment, where the 'best contracts' are reserved for the nicest or easiest recipients. In other words, the passage from theory to practice in the implementation of a 'contract' approach to social insertion requires a great deal of innovation, both in the social workers' profession (which was the underlying goal in France) and in the functioning of the local welfare system. At the same time, it is highly vulnerable to disappointment — among beneficiaries, as well as the community and policy makers — if contracts are not available or are not implemented. Open debates in social assistance discourses Income support measures for the poor have been under scrutiny in most countries for some time. Different issues are at stake, not only those concerning 'active' versus 'passive' measures. We point out three crucial sets of problems'5. The first concerns the criteria for defining the groups entitled to income support: to what degree should universalism be corrected through some form of selectivity based on age, residence, family responsibility, family dependence, and so forth? This involves several different problems. One concerns the degree to which family and kin solidarity should be enforced, particularly with regard to parent-children obligations. Another concerns the social and moral acceptability, or opportunity, of giving income support to the young. These two questions are linked, but also distinct. The latter implies not only moral judgements on family obligations, but also on the status of the young as citizens, on what society owes them and their specific vulnerability not only to poverty, but to dependence and passivity. In this respect, it is interesting to consider the rules under which some countries — for example, France, Luxembourg, Spain and Denmark — exclude young people below a certain age from minimum income schemes. It would be too simplistic to interpret it solely in terms of a denial of rights. This choice points to a view of the specific moral vulnerability of the young, and also to the definition of what should be offered to them as an alternative: training, work experience, and so forthy. On the other 251 Social assistance dynamics in Europe hand, in Denmark (and Sweden since January 1998), young people (18-24 years old) applying for income support have been obliged since 1990 not only to register at the unemployment office, but to accept any training or job offered to them by the municipality, under penalty of losing the unemployment benefit. Although this system may appear strongly compulsive, it still grants the young more resources than in France or Spain (not to mention Italy), where a young person may or may not be offered a training programme, or job experience, while being excluded from the minimum income scheme. Yet another probleríí concerns the entitlement of non-nationals, and particularly Third World immigrants. To what degree and under what conditions (duration and legality of residence, past contributory or tax paying history), should they be entitled to sharing the national or local welfare resources? One way of addressing (or solving) the issue of entitlement is that of prioritisation (and/or categorisation), which is also a way of defining the deserving or undeserving poor. Extreme categorisation, both at the national and local level, occurs in countries such as Italy where there is no nationally regulated income support scheme. But also in countries where there is a national scheme, as in Prance, Portugal, Spain, it may be added to other existing categorical schemes (for older people, people with disabilities, lone mothers, widows, or poor working adults) as a last resort. In any case, the issue of who deserves or is entitled to social assistance tends to be periodically reformulated according to criteria that may also be dictated by political cultures and expediency. A case in point is the different way of framing the needs and characteristics of lone mothers in different countries: while in some they may be considered the most deserving category (together with older people), in others they are perceived as socially irresponsible users of the system, to be controlled and kept in check. A particular problem that arises within the issue of entitlement is that of the possible negative effects, particularly with regards to women, set in motion by means-testing of households. Not only is there a risk that the poverty of financially dependent women is not acknowledged through this approach, but their attempts to become financially autonomous may be thwarted. This is a particularly difficult, if not intractable problem, in so far as all countries, even the most individualistic in their approach to social protection, assume for social assistance purposes that spouses and cohabitant couples share their income. The second set of problems concerns whether income support should be seen primarily as a preventive or as a curative measure. In other 252 Deconstructing the myth of welfare dependence words, should social assistance1 act as a sort of launch pad, that is an incentive structure to prevent the socially vulnerable becoming poor or socially excluded, or as a last resort or 'safety net' for those who cannot help themselves. Though conceptually clear, this issue is easily confused at the policy level — not only because the same measure might be used in a preventive rather than curative (or even exclusionary) way, but because the time-horizon of preventive measures may be out of scale with that perceived and experienced by the poor themselves. Thus, work requirements or incentives may be empowering opportunities, pushing an individual towards adequate job insertion or, on the contrary, they may be used as a means of social insertion for individuals who are not (or no longer) employable, simply helping them to feel useful. At the same time, the timing of programmes — duration of benefits, training courses, work experience and so forth — may be out of step with the subjective perceptions of beneficiaries and with their personal and family circumstances. It should be added that the focus on labour market participation and paid work as the main route out of poverty, is highlighting not only the hypothetical disincentive effects of social assistance, but also the plight of the working poor. Thus, in a few countries policies are being developed specifically for this population, and particularly families with children (see also OECD, 1999). Once again, following and improving on the US experience with the Earned Income Tax Credit, the UK has been the first European country to develop such an approach, through the Working Families' Tax Credit. It is now being followed in France with a similar measure, and also has an equivalent in Ireland. Debate on how to support poor working families is going on in many countries. To be more precise, there are two debates, which should not be confused: one concerns income support for families who are below a given earnings' ceiling, the other concerns the subsidy of low productivity jobs. While the former involves integrating the inadequate income of a household, the latter involves subsidising (for example, by reducing the cost of labour) labour market demands for unskilled, or labour intensive jobs. The idea is that these provide employment for those actual or potential social assistance beneficiaries who have the most difficulty in becoming financially independent because of their low skills or other shortcomings. This is not the place to address the latter issue. As for the former, we should like to point out that, important as it is to acknowledge that efforts must be made 'to make work pay', the mechanisms involved in means-testing on a household basis risks creating further vicious circles, 9C3 Social assistance dynamics in Europe particularly for women. It can make it no longer worthwhile to work extra hours or to have an additional worker/earner in the household. This negative incentive affects in particular two-parent households, rendering them (and especially women) more, not less vulnerable to poverty and social exclusion should something happen to the couples relationship1". Paradoxically, this occurs in the same country, the UK, in which the New Deal for Lone Parents strongly encourages lone mothers to take up paid work as a way of better protecting themselves and their children financially, and being better integrated socially. In any case, we might suggest that 'making work pay' policies represent an institutionalisation - with all the advantages not only of higher generosity, but of legality and of some degree of social security - of the kind of'assisted equilibrium' achieved in some of the Southern European cities through the combination of inadequate income support and recourse to the informal labour market. Further, in the case of institutionalised making work pay policies, beneficiaries holding a job are not perceived as cheaters who might or might not be condoned on the basis of social workers' leniency or understanding. On the contrary, they are perceived as better citizens because they are less dependent on public assistance. The focus on 'active' policies also reformulates the issue of the balance struck between universalism and individually adapted measures. Special incentives (or disincentives), insertion programmes, requirements and 'contracts', cannot be overly standardised if they are to serve their purpose, as they must match the circumstances of beneficiaries. This opens up the whole question of the kind of obligations (both for the social services and for recipients) that means-tested income support for the able-bodied should involve, and represents the third issue in the current policy debate. What balance should there be between the respect of individual choice and personal freedom, and an authoritative approach that predefines the range of options and imposes a set behaviour? For example, the requirement of willingness to actively search for a job, which is imposed on able-bodied income support recipients in all countries, can represent a positive incentive - a means of achieving social integration and developing personal skills. Vice versa, it may also be a constraint that prevents recipients from developing alternative strategies and ways out of poverty. It can be used as an empowering device or as a constraint, aimed at getting people off the welfare rolls, at stigmatising them or even making it difficult for people to receive social assistance. Moreover, it can easily become an enforced obligation to work, turning social assistance beneficiaries into forced labour, in a contemporary reinvention of the workhouse". It is not at all easy to judge beforehand how this requirement 254 Deconstructing the myth of welfare dependence will be implemented and with what consequences. There is a fine dividing line between the two approaches — a line that is often crossed in both directions (more in the actual implementation of policies at the everyday level, than in formal rules and principles of law). Moreover, the risk of the avoidance of responsibility by social services in the name of freedom for the recipients is as great as that of over control and authoritarianism. It is interesting, from the point of view of the history of ideas and policy models, to view the different status of the term 'workfare' in the past three decades. It was developed in the US in the 1970s in criticism of work requirements introduced as a form of social control of the poor; thus it represented diminished citizenship rights. It has become the catchword of'progressive' social assistance policy both in the US and in Europe in the mid-1990s, though with a slight linguistic variation in the new UK New Labour vocabulary: 'welfare to work'12. This radical shift in usage and meaning certainly presents the risk of being translated into forms of Stigmatisation of social assistance recipients, if not forced labour. Yet it should not be interpreted merely as a form of political cynicism. Rather, it points on the one hand to the, largely unfounded, but still quite widespread fear of long-term dependence on social assistance. In some Southern European countries, with large informal economy sectors, this fear may be compounded by that oi condoning and even encouraging both fraudulence in obtaining benefits and participation in the illegal (and in the case of Italy, sometimes even criminal) economy. On the other hand, that language shift points to the awareness that inclusion must also address issues of empowerment, acknowledgement and strengthening of individual capabilities. Whatever the implicit and explicit goals, the stress on active measures also points to new roles and responsibilities for local actors, and particularly for local communities. These can no longer be perceived simply as a passive context in which the events of economic and social history have led to various forms of poverty and social exclusion being experienced. The community also consists of actors responsible for implementing policies — not only applying rules and regulations, but providing insertion programmes, developing 'contracts', forming partnerships, and so forth. The new active policies against poverty and social exclusion imply the existence of a local community: as a place in which poverty and/or exclusion arise, but also one in which forms of social integration are developed. From this point of view, specific sets of local actors can encourage both diversity and a degree of discretion in implementing policies. They can be the instruments for (re-)creating or (re-)structuring local communities, and thus possibly reinforcing diversity and inequality 255 Social assistance dynamics in Europe between communities, especially when available resources are limited. Consequently, diversity with regard to measures to combat poverty and social exclusion is likely to increase not only between but also within countries. This process of policy re-evaluation is going on with varying intensity in all European countries, as well as in other OECD states. Thus it includes countries that already have an established and well defined policy against poverty and some kind of minimum guarantee of resources, as well as countries that do not - at least not as an institutional part of the social rights package or of society's obligations and responsibilities in defining citizenship within the national (central or federal) community. From this point of view, comparative research allows us to understand how different systems define and deal with these and related issues, and what the results are. At the same time, comparative research can contribute both empirically and methodologically to the process of European policy making. Social exclusion/inclusion is one of the areas in which a common approach is being developed through the method of open coordination. In fact, in order to develop a common understanding of the goals of social policy and to benefit from mutual learning and best practice, it is important to have a clear understanding of the complex network of established practices and institutions, of specific cultural meanings and the patterns of sociability that underlie given policies. Policy transfer between countries is not easy. This is not only because individual policies are usually part of a far more complex package, but also because they reflect nation-specific ways of understanding reality. The opening of a European discourse, with a set of common goals and indicators, is an ambitious attempt to develop shared understandings — requiring more or less radical change in existing nation-specific and local patterns of discourse and policy. Careful testing of stereotypical views and nation-specific experience, contrasting modes of interpretations and patterns of intervention, is needed in order for this effort to be successful in terms of efficacious policy decisions. Notes 1 Literature on this phenomenon is substantial. See for example, Barbagli and Saraceno, 1998; Garfinkel and McLanahan, 1990; Marsiglio, 1995; Martin, 1997; OECD, 1990; Seltzer, 1994. 2 In the new edition of his retrospective longitudinal research on social assistance recipients in a small French town, Paugam (1997b) stresses that different groups 256 Deconstructing the myth of welfare dependence might also represent different stages in the life course of social recipients, although not all recipients go through all phases. 3 See for example, Grover and Stewart, 1999; Jessop, 1993; Standing, 1990. The studies collected in the European Foundation for the Improvement oi Living and Working Conditions, 1999;L0demel andTrickey, 2000. "* As a consequence of the Lisbon and Nice summits in 2000, social exclusion has entered the European policy agenda and countries are requested to prepare two year National Action Plans on the basis of common objectives. Analogous to the National Action Plans on employment, they will be assessed through a peer review process. 5 The role of the third sector has been one of the focuses of a TSER funded project on the 'Evaluation of local socio-economic strategies in disadvantaged urban areas', ELSES (project 3047, coordinators S.Weck and R. Zimmer Hegman). See ELSES, 2000. '' Abrahmson (2000) advances this observation for the Danish case, pointing out that jobs or training courses offered through the social assistance channel are less likely to lead to regular jobs than temporary jobs or training experience offered by the non-protected labour market and training system, since the former are stigmatised. 'Jordan (1996, p 36) argues that for people trapped in deprived 'communities of fate', illegality may be a more secure source of income for survival than the new flexible labour market, where extremely irregular employment for below subsistence wages does not provide the basis for a sustainable way of life. H See also Leisering and Walker (1998b), whose list is slightly different, and more articulated, than ours. 9 In France, the Emploijeiines programme — a voluntary scheme, launched in 1997 — was the alternative answer to the request that the age threshold for obtaining the RMI be lifted. 10 We are not denying that the Working Families'Tax Credit is an improvement compared to the previous limitation in working hours allowed, in so far as it addresses the plight of poor dual-parent working households (see for example, Piachaud and Sutherland, 2000). Yet, the effects on the behaviour of all household members in the way this support is given, and its implicit assumptions concerning 257 Social assistance dynamics in Europe the gender division of labour, should not be underestimated. Further, a more generous acknowledgement of the cost of raising children by providing increased child benefits and/or tax allowances could possibly have the same protective effect without creating disincentives. 1' See for instance, Sunesson et al (1998), who argue that this is presently a risk in Sweden. See also Becker, 1997. 12 There is not a general consensus on what workfare includes and the use of the term varies over time and across countries (Peck, 1998), besides being a highly charged political term. A restricted definition, used also in a recent comparative study of workfare policies in six European countries, defines it as "policies which require people to work in exchange for, or instead of, social assistance benefits". See Lodemel andTrickey, 2000. For an overview of debates and experiences, see European Foundation for the Improvement of Living and Working Conditions, 1999. See also Barbier, 1998; Standing, 1990. 258 erence: Abrahmson, P. (2000) 'L'activation des politiques sociales scandinaves: le cas du Danemark', in C. Daniel, and B. Palier (eds) La protection sociale en Europe, Paris: Ministěre de ľEmploi et de la solidarite, Mission recherche-DREES,pp 123-40. Alber, J. (1995) 'A framework for the comparative study of social services', Journal of European Social Policy, vol 5, no 2, pp 131-49. Allmendinger, J., Brückner, H. and Brückner, E. (1993) 'The production of gender disparities over the life course and their effects in old age: results from the West German Life History Study', in A.B.Atkinson and M. Rein (eds) Age, work and social security, New York, NY: St Martin Press, pp 188-223. Amrouni, I. and Math,A. (1997/98) 'Une part importante des beneficiaries de minima sociaux les percoit ä un taux réduiť, Rechcrches et Previsions, no 50/51, pp 49-67. Andreß, HJ. (ed) (1998) Empirical poverty research in a comparative perspective, Aldershot: Ashgate. Andreß, H.J. and Schultz, K. (1998) 'Poverty risks and the life cycle: the individualization thesis reconsidered', in HJ. Andreß (ed) Empirical poverty research in comparative perspective, Aldershot: Ashgate, pp 331-56. Ashworth, K., Hill, M. and Walker, R. (1992) Patterns of childhood poverty: New challenges for policy, Working Paper no 169, Loughborough: Centre for Research in Social Policy, University of Technology. Astier, I. (1997) Revenu minimum et souci ď insertion, Paris: Desclée de Brouwer. Atkinson, A.B. (1998a) Poverty in Europe, Oxford: Basil Blackwell. Atkinson, A.B. (1998b) 'Social exclusion, poverty and unemployment', in A.B. Atkinson, and J. Hills (eds) Exclusion, employment and opportunity, CASE Paper 4, London: London School of Economics, pp 1-20. AubertjB. (1997/98) 'Seuils de pauvreté et montanes des minima sociaux. Remarques sur les usages et les discours', Rccherches et Previsions, no 50/ 51, pp 69-80. 259 Article Poverty in Europe in the mid-1990s: the effectiveness of means-tested benefits Diane Sainsbury, University of Stockholm, Sweden Ann Moris sens'1", Roskilde University, Denmark Summary This article examines the income maintenance policies of several members of the European Union and three candidate countries: the Czech Republic, Hungary and Poland. It addresses the issue of the effectiveness of these policies and especially means-tested safety nets .-'in alleviating poverty. To assess the effectiveness of the policies, we use data from the Luxembourg Income Study. We analyse the incidence of poverty based on the EU poverty line and poverty reduction for the entire population and vulnerable groups - the unemployed, solo mothers, large families, and the elderly. During the 1990s die poverty rates increased in most countries and for most vulnerable groups. Means-tested benefits assumed growing importance in alleviating poverty, and several countries have improved their schemes to guarantee a minimum income. At the same time reforms have produced diversity in the safety nets across Europe. Key words European Union, means-tested benefits, poverty, safety nets, social protection Résumé Cet article examine les politiques de maintien des revenus au sein de plusieurs états membres de ľUnion Européenne et de trois pays candidats: la République Tchěque, la Hongrie et la Pologne. U s'intéresse ä ľefficacité de ces politiques et en particulier des minimums soci-aux sur la reduction de la pauvreté. Pour évaluer ľefficacité de ces politiques, nous utilisons les données récoltées par l'Etude sur les Revenus de Luxembourg (Luxembourg Income Study). Nous anaiysons les situations de pauvreté sur base de la ligne de pauvreté définie par PUE et la reduction de la pauvreté dans la population dans son ensemble et au sein des groupes plus vulnerables - les chômeurs, les méres céli-bataires, les families nombreuses et les person-nes ägées. Au cours des années 1990, le taux de pauvreté a augmente dans la plupart des pays et pour la plupart des groupes vulnerables. Les allocations sous conditions de ressources ont eu un role croissant dans la reduction de la pauvreté et plusieurs pays ont amélioré leurs dis-positifs pour garantir un revenu mimimum. En méme temps, les réformes ont conduit ä une diversité des filets de sécurité en Europe. • Introduction Combating poverty has been an official concern of the Commission and the member states since the mid-1970s when the first Poverty Programme was adopted.1 The poverty issue assumed new urgency with the slowing of economic growth and rising unemployment in the 1980s. Around 1990 nearly 50m Europeans were living under the poverty line (Eurostat, 1994: 185) and towards the end of the decade the number had risen to 60m Europeans (COM/2001/565: 6). The 1992 Recommendation on Sufficient Resources called for the establishment of common criteria concerning sufficient resources and social assistance in the social protection systems of the member countries (Abrahamson, 1997: 140). More recently, in 'A Concerted Strategy for Modernising Social * Author to whom correspondence should be sent: Ann Morissens, Department of Social Science, Postboks 260, 4000 Roskilde, Denmark, [email: annmo@ruc.dk] Journal of European Social Policy 0958-9287 (200211)12:4 Copyright © 2002 SAGE Publications, London, Thousand fUUc nr,A M«.,. n.lk; lHnHum an-?, mocno 308 Sainsbury and Morissens Protection1 (1999), the Commission set as a goal to 'ensure effective safety nets, consisting of minimum income benefits_and accompany^ Ihg provisions, with a view to efficiently combat poverty and exclusion of individuals and families' (COM/1999/347). This article examines the income maintenance policies of several European countries and their role in combating poverty, and it addresses three issues. First and foremost, how effective are means-tested safety nets and other social transfers in alleviating poverty? Second, has there been a convergence in the safety nets of the member countries of the European Union during the 1990s? Third, what are the implications of enlargement of the European Union for the creation of a common safety net? To answer these questions we use two waves of the Luxembourg Income Study from the early and mid-1990s to analyse the income maintenance policies of several member countries and three Central European countries that are candidates for membership. We have selected countries that represent different welfare regimes. The member countries include Belgium, France, Germany, the Netherlands (representing the Conservative Corporatist regime), Italy, Spain (the Southern European regime), the United Kingdom (the Liberal regime), Denmark, Finland and Sweden (the Social Democratic regime).2 The candidate countries are the Czech Republic, Hungary and Poland whose prospects of membership have been the most promising. These countries share a Soviet legacy and the transitional experience, leading to assumptions of a common welfare regime quite different from those of the other European countries. Already in die early 1990s, however, it was suggested that the emerging post-communist welfare state regimes of Czechoslovakia (especially the Czech lands), Hungary and Poland represented three distinctive types: the Social Democratic, the Liberal Capitalist and the Post-communist Conservative Corporatist respectively (Deacon, 1992: 172-83). First we discuss how our analysis differs from earlier studies. After presenting the data and methods, we measure poverty for the jrjtir^__pcipulation,__along—with—vulnerable-groups, across countries. We examine the effectiveness of income maintenance policies in reducing poverty, distinguishing between the impact of means-tested benefits and other social transfers. In the concluding discussion wě comment on trends of convergence and the implications of EU enlargement for the prospects of a European safety net. Previous research Until recently, the study of means-tested benefits has suffered from two major weaknesses. First, relatively little scholarly attention had been devoted to means-tested benefits. Instead this type of benefit has been overshadowed by research on social insurance schemes. Second, there has been a general tendency to assume that means-tested benefits are characterized by invariance. Gesta Esping-Andersen's influential book The Three Worlds of Welfare Capitalism (1990) has also contributed to diverting attention away from analysing variations in means-tested benefits. His regime typolog)'- assigns special importance to means testing and makes it a defining characteristic of the Liberal welfare state regime. In the process he obscures the existence of means-tested benefits and their performance in the other welfare state regimes. In addition, his emphasis on decommodification, which presupposes commodification, underlines the importance of work-related benefits. To date the most comprehensive study of variations in means-tested benefits is Social Assistance in the OECD Countries (Eardley et al., 1996), summarized in this journal (Gough et al., 1997). The study highlights structural dimensions of variation, such as the position of means-tested benefits and linked benefits within the social security system, eligibility criteria and resulting coverage, administration and benefit levels (Gough et al., 1997).3 In contrast, we examine the functioning of Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 309 means-tested programmes rather than their structural variations. Of central significancejn^ llTe analysisliere is the dimension of poverty-reduction effectiveness and the restating poverty rates. Furthermore, we are interested in the poverty-reduction effectiveness of means-tested benefits in relation to other social transfers. We define the basic safety net as the range of benefits available to guarantee a minimum income based on a resource test (cf. Eardley et al., 1996: 1). According to this definition the safety net consists of all means-tested benefits in a particular country regardless of whether the resource tesť is a poverty test to restrict access to the most needy or an affluence test to exclude the well-off. We prefer not to equate the safety net with" social assistance,, which could result in omitting other means-tested benefits that are separately administered from assistance schemes. Such omissions rest on an administrative and not a theoretical distinction (cf. Gough, 1996). Moreover, emphasis on social assistance in the definition of the basic safety net may block consideration of new ideas and innovative programmes, such as a participation income (Atkinson, 1998: 145-9), the Finnish reform of unemployment benefits (OECD, 1998a: 70) or the Czech state social subsidies system to guarantee a minimum living standard (OECD, 19?8bri76-84). Data and methods For the analyses presented in this article we use two waves of data - collected in the early and mid-1990s - from the Luxembourg Income Study (see the Appendix Table 1, for a list of the years and surveys). The Luxembourg Income Study (LIS) is a cross-national data set with detailed information about sources of household income in over 20 countries. A special feature of LIS is the harmonization of the main income variables -disposable income, gross income and market income - which creates a unique data bank for researchers to conduct cross-national _anaiyses-of^pover-ty^ÍHGome-distr-ibution-and inequality. Despite the efforts to harmonize the income variables, problems of data comparability have not been fully resolved. In this article we have confronted three problems. The first concerns differences in the content of the means-tested benefits variable. Basically the LIS means-tested benefits variable includes social assistance benefits and housing allowances. Our definition of the safety net includes all types of benefits that entail some sort of resource test and therefore requires a recoding of the LIS means-tested variable for some countries (see Appendix for details). The second limitation is that the LIS does not contain information on benefits in kind. Accordingly our analysis is confined to income poverty. Lastly, because of data limitations we do not analyse tax benefits targeted to low-income groups.4 Our approach differs from methods that have figured prominently in earlier research on two counts. First, much of die recent Eurgpean^literature has assessed social assistance employing a model recipients approach (e.g. SŽW7l995fBfadshaw etal.,' 1996: chs 5-é; Eardley et al., I996j Cough et al., 1997). This approach focuses on the statutory provision of benefits and the entitlements of typical but hypothetical recipients with specific traits; it assumes that the model recipients claim and receive the benefits to which they are entitled. Although useful in identifying and comparing the basic features of social programmes, the approach is often not very informative about programme outcomes. A major weakness is that it provides a picture of how social provision could or shoidd work but not how it actually does work. An additional difficulty concerns the representative trade-off of the model recipients approach (Eardley et al., 1996: 8, 116). As the models become more elaborate and supposedly more realistic, problems of representativeness mount because of the growing number of specific assumptions made about recipients. Instead of model recipients we use the LIS, which allows 310 Sainsbury and Morissens us to examine real recipients and actual outcomes.5 Second, we propose a new measure to -Ga-pture-the-impaGt-Gf-means-tesded-benefits-on— the economic situation of persons with low incomes. The most common method to assess the impact of means-tested benefits, has .been, to analyse their contribution to income. One procedure has been to examine the income structure across income groups from the poorest to die wealthiest (quintiles or deciles) to determine the contribution of these benefits to total income for each quintüe or decile (e.g. Deleeck et al., 1992). Another tactic has been to analyse the income packages of particular . groups - families, solo mothers, etc. (Rainwater et ah, 1986). The income package is a sum of income acquired from several different sources. While this concept provides a tool to break down income by its sources to determine the importance of each component, a shortcoming is that quite similar income packages can be associated with very different poverty rates.6 Rather than emphasizing the contribution of means-tested benefits to income, we are interested in whether these benefits make a difference by altering one's poverty status. We focus on the extent to which the receipt of means-tested benefits lifts a person over the poverty line. Poverty incidence We use the relative approach in measuring income poverty. This means that we define as poor those households that have a disposable income adjusted for family size - or an equivalent disposable income - below a certain threshold representing the level of well-being of the population in a specific country. In most comparative studies the poverty threshold has been set at 50 percent of median equivalent disposable income. Instead we use the EU's current definition of poverty - 60^ percent of median disposable income adjusted for family size (Eurostat, 2000). To adjust disposable income for family size we use the OECD equivalence scale. This scale assumes that a child under 18 needs 50 percent of what an adult needs and that each additional adult -in-the-househofdmeeds^^ first adult in the household needs. This equivalence scale is also used by the EU. Poverty-reduction effectiveness To assess the effectiveness of public social transfers in general we compare the population's economic situation at two points -before and after taxes and transfers - and use relative and absolute measures of poverty reduction. The relative effectiveness of poverty reduction is measured as follows: Pre-transfers and taxes poverty rate - Post-transfers and taxes poverty rate/Pre-transfers and taxes poverty rate * 100 (Mitchell, 1991: 65). We use the poverty rates based on market income as a measure for the pre transfers; and taxes situation.,,Market income includes earnings from labour and capital as well as occupational and private benefits. Poverty rates based on equivalent disposable income are used as a measure for after transfers and taxes. In assessing the relative effectiveness of means-tested benefits in alleviating poverty we employ a similar measure. Now we compare the poverty rates before and after means-tested benefits (for the exact content of means-tested benefits, see the Appendix). More precisely, we use poverty rates based on disposable income minus, means-tegted benefits and compare them with poverty rates based on the disposable income. The former is the poverty rate based on the income people have after taxes and transfers before receiving means-tested benefits, whereas the latter includes means-tested benefits and we can attribute the reduction in poverty to means-tested benefits. Here the relative effectiveness of poverty reduction is measured in the following way: Pre-means-tested poverty rate -Post-means-tested poverty rate/Pre-means-tested benefits poverty rate * 100. Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 311 We also utilize absolute measures of poverty-reduction effectiveness. More generally, the absolute measure is the percentage, point difference between the poverty rate before and after transfers and taxes. In determining the impact of the safety net, the absolute measure refers to the percentage point difference in the poverty rate before and after means-tested benefits. Finally, it needs to be stressed that our measures of poverty rates and poverty reduction are after taxes.7 Poverty and poverty reduction: social transfers and" the safety net First the poverty rates for the entire population in the 13 countries are presented. We then examine poverty reduction achieved through all social transfers and means-tested benefits in particular. Subsequently we look at vulnerable groups - the unemployed, single parents, large families, and the elderly - and the impact of social transfers and the safety net. Poverty rates Figures la and lb present the poverty rates of the national population in the early and mid-1990s, using the EU's current definition of poverty - 60 percent of median disposable income adjusted for family size. At both points in time the poverty rates of the Czech Republic stand out as the lowest. In the early 1990s the Czech Republic was followed by Denmark, Finland, the Netherlands, Belgium, Germany, Hungary and Sweden. The remaining countries - Poland, Spain, France, Italy and the UK - had higher poverty rates that were above the mean. In the mid-1990s there was some reshuffling among countries within the groups but little movement between clusters. Primarily Hungary moved from being a country whose poverty rate was below the mean to one above it. The pattern of clustering based on poverty rates is quite different from that of welfare state regimes, especially in the early 1990s. Among the countries with low poverty rates _we__find—representatives—of—tire—Suciat Democratic regime and the Conservative Corporatist regime as well as member and candidate countries. Likewise the nations with higher rates of poverty represent several regime types and both members of the EU and candidates for membership. However, countries representing the Social Democratic regime did not have high poverty rates and those representing the Southern European regime did not have low poverty rates. It should also be noted that overall the national poverty rates increased during the first half of the 1990s. This is evident in the poverty rates of the individual countries and the mean. The two exceptions were France and Sweden. The most dramatic increase occurred in the Czech Republic, although its poverty rate remained strikingly low. Notable increases also took place in the other two transition countries. Furthermore, in the mid-1990s the poverty headcounts of three countries - Italy, Poland and the UK -exceeded 20 percent compared to only the UK at the beginning of the decade. Poverty reduction Table 1 presents relative and absolute measures of poverty-reduction effectiveness for all social transfers. The relative measure of poverty-reduction effectiveness indicates the proportion of pre-transfer poor lifted above the poverty line through social transfers. The absolute measure is the percentage point difference between the pre-transfer and post-transfer poverty rates. In comparing the absolute measure across countries, it ought to be noted that its magnitude is limited by the pre-transfer poverty rate. The table shows pronounced differences in the performance of the social protection systems of the countries in reducing poverty both in relative and absolute terms. This was especially true in the early 1990s when the Journal of European Social Policy 2002 12 (4) 312 Saixisbury and Morissens Mean CZ92 DK92 FI91 NL91 BE92 GEB9 HU91 SW92 PL92 SP90 FRB9 IT91 UK91 Country/year Figure la National poverty rates (early 1990s) Source: LIS; own calculations. Mean CZ96 SW9S FI94 DK95 NL94 BE97 GE94 HU94 FR94 SP95 UK95 PL95 IT95 Country/year Figure lb National poverty rates (mid-1990s) Source: LIS; own calculations. Czech system lifted 9 out of 10 of the pre- as the British system improved and the effec- transfer poor above the poverty line in con- tiveness of Czech transfers declined. The rela- trast to approximately 3 out of 10 in the UK. tive effectiveness scores of many countries also In lie mid-1990s divergence was not so sharp, decreased during the decade. Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 313 Table 1 Rank order of countries based on poverty-reduction effectiveness of all social transfers for entire population ____ Early 1990s Czech Republic (1992) 90.7 Hungary (1991) 75.1 Sweden (1992) 70.7 Poland (1992) 68.5 Denmark (1992) 68.1 Belgium (1992) 65.4 France(1989) 59.8 Netherlands (1991) 56.2 Finland (1991) 55.4 Spain (1990) 54.4 Germany (1989). 52.9 Italy (1991) 48.7 UK (1991) 34.4 Mid-1990s Czech Republic (1996) 76.8 Sweden (1995) 75.1 Hungary (1994) 70.3 France(1994) 63.2 Poland (1995) 61.8 Finland (1995) 60.8 Belgium (1997) 60.5 Germany (1994) 53.6 Netherlands (1994) 51.2 Italy (1995) 46.1 UK (1995) 41.3 Relative-poverty^reduction----------Absalutepoverty^redírctiofr 28.2 35.6 28.8 36.3 19.5 21.2 27.1 14.2 11.7 20.1 13.3 18.6 10.8 21.5 29.2 38.2 29.4 36.5 15.9 20.5 15.7 14.0 19.9 14.5 Source: LIS; own calculations. More specifically, the Czech Republic not only distinguished itself by its low poverty rate but also its high relative effectiveness scores at both points in time. The other two transition countries - Poland and Hungary -displayed very high levels of market income poverty, high scores of relative effectiveness, and simultaneously the greatest absolute poverty reduction through social transfers. However, since the levels of market income poverty were so high, their poverty rates after transfers remained high. Sweden and Denmark were also included in the cluster with high relative poverty-reduction scores, and the two countries had low poverty rates. (Unfortunately we do not have reliable Danish data for the mid-1990s.) Finally, Italy and the UK exhibited the lowest poverty reduction in absolute terms and the highest poverty rates; and they also had the lowest relative poverty-reduction scores. What is the role of the social safety net in poverty reduction? Table 2 presents the rank order and clustering of countries on the basis of poverty reduction by means-tested benefits for the entire population. The relative poverty-reduction scores here refer to the percentage of the poor prior to means-tested benefits lifted out of poverty. The absolute score is the difference in the poverty rate before and after means-tested benefits. For both waves of data, several similarities in the pattern . of clustering are observable,8 but because of the increasing importance of means-tested benefits in the relative reduction of poverty there is an upward shift with countries moving into the next cluster. journal of European Social Policy 2002 12 (4) 314 Sainsbury and Morissens Table 2 Rank order of countries based on poverty-reduction effectiveness of means-tested benefits for entire population -Relativepovertyrrpdirction Ab soluWp overly reduction Early 1990s Sweden (1992) 38.0 7.2 Denmark (1992) 37.5 5.5 Finland (1991) 28.7 3.8 UK (1991) 20.0 5.2 Netherlands (1991) 17.1 2.3 Czech Republic (1992) 14.0 0.5, Hungary (1991) 8.2 1.1 Germany (1989) 7.5 1.0 Spain (1990) 4.6 0.8 France (1989) 3.2 0.6 Belgium (1992) 1.0 0.2 Italy (1991)a 0.0 0.0 Poland (1992)a 0.0 0.0 Mid-1990s ; Czech Republic (1996) 39.4 4.2 Sweden (1995) 36.2 5.5 Finland (1995) 33.1 5.1 UK (1995) 29.1 8.5 Germany (1994) 16.7 2.7 Netherlands (1994) 16.2 2.6 Poland (1995) .15.2 4.0 Hungary (1994) "9.7 1.8 France (1994) '9.1 1.7 Belgium (1997) 5.4 0.8 Italy (1995) 0.7 0.2 Note:a There is no means-tested benefits variable in datasets for Italy 1991 and Poland 1992. Source: LIS; own calculations. Looking first at the group characterized by benefits. To a large extent, this cluster was disminimal poverty reduction, the relative scores tinguished by the combination of high poverty are around 5 percent or lower, and the coun- rates after insurance-based social transfers and tries making up the cluster represent die ineffectual safety nets. Belgium is an exception Conservative Corporatist and Southern widi regard to the entire population but, as European welfare state regimes. A shared we shall see, fits this description with respect feature of these countries is that liability for to specific vulnerable groups - the unem-social assistance extends beyond the nuclear ployed and the elderly. family. In the early 1990s the cluster also The intermediate group (scores over 5 included Poland whose social insurance policy percent and under 20 percent) was made up of legacy bares a strong resemblance to the other Germany, Hungary, the Czech Republic and countries. Typical of the Corporatist regime, the Netherlands in the early 1990s. At that Polish insurance benefits have been frag- time the social provision system of the coun- mented and differentiated by economic sector. tries, with the exception of Germany, had a Nearly all the countries had rudimentary strong universalist thrust. The countries had social assistance programmes administered at relatively low poverty levels after social insur- the regional or local level, and all offered low ance and other transfers, and their safety nets Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid^l990s 315 made a moderate contribution to reducing poverty levels. In the mid-1990s Poland and "France joinedlfhis^clüster. The third group consists of the Scandinavian countries and the UK for both waves of data. The availability of social insurance benefits and/or non-contributory 'universal' benefits are often assumed to marginalize the utilization of social assistance and means-tested benefits. Accordingly, the universalism and prominence of other benefits in the Nordic social protection systems have prompted researchers to characterize the position of means-tested benefits as residual social assistance (Eardley'ét al., 1996: 169). Contrary to such categorizations, the means-tested benefits of the Nordic countries prove to be quite important in reducing poverty. These results also run counter to the picture of stringent administration and rigid asset tests for social assistance associated with Scandinavia (e.g. Bradshaw and Terum, 1997; Ledemel, 1997).9 Perhaps even more surprising is that their effectiveness matches or exceeds means-tested benefits in the UK. As we shall shortly see, for several vulnerable groups, means-tested benefits in the Nordic countries often have higher scores of relative poverty-reduction effectiveness than British benefits. Upon further reflection, however, the results are actually not so surprising. A relatively large share of the population claimed means-tested benefits in the Nordic countries, although not as large á share as in the UK, but Nordic benefits on the whole were much more generous than British benefits (Gough et al., 1997: Tables 2 and 6). Vulnerable groups and poverty reduction Now we turn to groups typically over-represented among the poor - the vulnerable for whom the safety net is supposed to guarantee a minimum income. We are particularly interested in how the safety net and other social transfers affect their poverty status. For each of the vulnerable groups, we report their poverty rates and then assess the impact of transfers, and especially means-tested benefits, on their poverty rates. The unemployed The unemployed are the most vulnerable in the sense that they on average exhibit the highest poverty rates of the groups we discuss.10 Their poverty rates ranged from a low of 15 percent (Denmark) to over 80 percent (Italy) in the early 1990s, and between 20 percent (Germany) and over 70 percent (Italy) in the mid-1990s (Table 3). As the decade progressed, the risk of poverty for the unemployed increased; in half the countries the unemployed had a poverty rate of over 50 percent in the mid-1990s. Of the candidate countries, Hungary had the lowest poverty rates, and they were lower than those of several EU members. Figure 2 summarizes the importance of the safety net and other social transfers in alleviating poverty among the unemployed. The height of die bars indicates the pre-transfer poverty rate of the unemployed. The figure shows the absolute poverty reduction and the resulting poverty rate; it also breaks down the contributions of the safety net and other social transfers to poverty reduction. Table 3 Unemployed under the EU poverty line (%) Early 1990s Mid-1990s Denmark 15.0 — Finland 24.9 25.9 Hungary 26.8 36.5 Sweden 28.7 21.3 Poland 30.2 51.5 Czech Republic 37.8 63.1 . Netherlands 40.3 28.4 France 45.0 51.5. Germany 47.1 19.8 Belgium 48.1 55.4' Spain 66.3 — UK 70.8 64.7 Italy 82.6 72.4 Source: LIS; own calculations. Journal of European Social Policy 2002 12 (4) 316 Sains bury and Morissens DKS2 GE94 SW95 FI95 NL94 HU94 FR94 PL95 BE97 CZ96 UK95 SPSO IT95 Country/year Figure 2 Poverty reduction for unemployed (safety net and other transfers) Source: LIS; own calculations. There is wide variation in the contribution made by the safety net, from a meagre 1 percent (Italy) to 23 percent (Finland). In most of the countries other social transfers played a larger role than the safety net in poverty reduction of the unemployed; The exceptions were Finland, the Netherlands, the Czech Republic and the UK.) Contrary to expectations, it is not the UK, representing the liberal welfare state regime, but Finland and the Netherlands where means-tested benefits produced the largest reduction in the poverty rate. However, the UK is distinctly different because social assistance dwarfed insurance benefits. In the other three countries social transfers continued to play a significant role, and their significance approximated that of means-tested benefits. In Denmark and Sweden the safety net complemented other social transfers. These transfers substantially reduced the poverty rate, and means-tested benefits further lowered it. Means-tested benefits lifted between 40 percent and 50 percent of the poor after insurance benefits above the poverty line in Denmark, Finland, the Netherlands and Sweden. The relative poverty- reduction scores of British, Czech, Polish and Hungarian means-tested benefits indicate medium performance (10-25 percent). Despite this, the UK had one of the highest poverty rates for the unemployed because other social transfers did not protect the jobless. The safety net was of less importance in Germany, Spain, France, Belgium and Italy. Except for Germany where other social transfers effectively lower the poverty rate, the social protection system failed to keep nearly half or more of the unemployed out of poverty and the safety net made little difference. Equally^worrisome, the.capacity of other social transfers to lift the. unemployed above the poverty line was only in the range of around 3 out of TO in several countries -Finland, Belgium," the Netherlands, France, Hungary and Poland - and even lower in the Czech Republic and Italy. Of these countries, only Finland and the Netherlands responded by strengthening their safety nets for the unemployed. However, social transfers in Germany, Sweden and Denmark revealed a stronger capacity to lift the unemployed over the poverty line. Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 317 Solo parents and large families -Solo—mothers—and—their-^children^rrave-beeiT very vulnerable to poverty, but this does not necessarily have to be the case, as witnessed by their poverty rates (Table 4). In the early 1990s wide variation characterized solo mothers' poverty rates. Basically, the countries divided into two camps: low rates in the three Nordic countries, Belgium and the three candidate countries; high rates in the remaining coimtries - with the highest rates in the UK, Germany, France and the Netherlands. In the mid-1990s the poverty rates in the Nordic countries remained low; the major changes were that the economic well-being of solo mothers..deteriorated.in Italy, Belgium and the three transition .countries, especially the Czech Republic. Nonetheless, the poverty rates in Belgium and the three candidate countries were in the middle range, although the Czech Republic was. edging upwards towards the group with the highest poverty rates. The situation of large families with three or more children has been eclipsed by the attention given to solo mothers in recent years. These families have poverty rates that often rival or in some instances exceed those of single parents. In the mid-1990s large families had a higher poverty rate than solo mothers in Finland, Belgium, Italy, Hungary and Poland -"alíd^Sp^n^íntTířearly 1990s (Table 5). The importance of the safety net and other social transfers in reducing solo mothers' poverty rates is presented in Figure 3, using the same measures as for the unemployed in Figure 2. Overall, compared to the unemployed, solo mothers' pre transfer poverty rates were lower since many had earnings. The Netherlands, however, was an exception; and in the UK solo mothers' pre transfer poverty rate was nearly as high as that of the unemployed. As can be observed in Figure 3, social transfers and the safety net kept many solo mpthers and their children out of poverty. Indeed, there are several parallels in the poverty reduction of solo mothers and the unemployed. Again, variation in poverty reduction by the safety net is substantial - and even slightly greater than for the unemployed. Now poverty reduction ranges from a mere fraction (Italy and Hungary) to 28 percent (Netherlands). Besides the Netherlands, the countries whose safety nets markedly reduced the. poverty rates of solo mothers were the UK, Denmark, the Czech Republic, Poland, Sweden and Finland. The safety nets of these countries also had the highest relative poverty- Table 4 Solo mothers under the EU poverty line (%)_______________________________ Early 1990s Mid-1990s Table 5 Couples with 3+ children under the EU poverty line (%) _______________Early 1990s Mid-1990s Hungary 12.1 23.6 Czech Republic 6.8 20.ÍL. Finland 12.2 12.7 Sweden 16.8 1L9 Czech Republic 13.0 34.9.. Finland 18.3 22.6 Sweden 14.7 14.1 Denmark 18.8 — Denmark 16.2 — Netherlands 22.3 17.9 Poland 16.5 28.6 Belgium 23.4 24.9 Belgium 16.9 22.4 Germany 25.4 34.4 Italy 28.1 49.0 Hungary 27.5 34.4 Spain 33.0 — Poland 31.8 47.6 France 39.8 38.1 France 32.0 30.2 Netherlands 43.3 38.8 Spain 35.5 — Germany 44.9 52.6 UK 39.0 41.9 UK 53.8 52.2 Italy 47.0 51.8 Source: LIS; own calculations. Source: LIS; own calculations. Journal of European Social Po/iev 2002 12 (4) 318 Sainsbury and Morissens FI95 SW95 DK92 BE97 HU94 PL95 SP90 CZ9B FR94 NL94 IT95 UK95 GE94 Country/year Figure 3 Poverty reduction for solo mothers (safety net and other transfers) Source: LIS; own calculations. reduction scores. The Danish safety net lifted nearly 6 out of 10 pre-means-tested poor mothers above the poverty line; the Swedish, Finnish and Czech safety nets aided roughly 4 out of 10; and the British and Polish safety nets around 3 out of 10. The effects of the safety net in reducing the poverty rate were either limited or negligible in Belgium, France, Germany, Spain, Hungary and Italy. However, the safety nets in France, Germany and Belgium did pull between 1 and 2 solo mothers out of 10 over the poverty line. The impact of other social transfers on the poverty rate was considerable - a reduction, of between 20 and 50 percentage points - in Belgium, Poland, Hungary, Sweden, Finland, Spain and France. Thus in both Finland and Sweden other social transfers and the safety net combined to keep solo mothers and their children out of poverty. Among the differences - evident through a comparison of Figures 2 and 3 - is a gender bias in the protection system that is particularly sharp in Germany but also discernible in the Netherlands. The unemployed fared much better than solo mothers. Interestingly the pattern is reversed in Belgium, Poland, Spain, France, Italy, the UK and the Czech Republic. Turning to large families, we find that the safety net is important in Denmark, Finland, Sweden, the UK, the Czech Republic, Poland and Hungary, with the largest reduction in the poverty rate occurring in the Czech Republic, followed by Sweden and the ÜK. The relative effectiveness scores of means-tested benefits were much higher iii Sweden (32.9) and the Czech Republic (25.3) than in the other countries. (It is noteworthy that the safety net also aided families with less than three children in Finland, Sweden, the Czech Republic and the UK.) As distinct from the other countries, British means-tested benefits again accounted for nearly all the absolute poverty reduction. Other social transfers had a major impact on the poverty rate of large families, lowering it by at least 20 percentage points in Hungary, France, Belgium and Sweden. Neither means-tested benefits nor other transfers had much effect on the poverty rate of large families in Italy. Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 319 The elderly higher poverty rates than elderly (COM/2001/565: 169, 172). Tradition allythe elderly have been included as a vulnerable group because their economic well-being largely depends on the social protection system. Pension reforms and improvements in retirement income put in question the vulnerable status of the elderly. Nevertheless, our first wave of data from the early 1990s indicates that persons aged 65 years and older had a higher poverty rate than the national average in most of our countries (Table 6), while the second wave of data presents a much rosier picture. In the mid-1990s the poverty rate of the elderly was lower than the national average in nearly all the countries, and it had fallen in all the countries except the Netherlands, Belgium and the Czech Republic. However, data from the recent Draft Joint Report on Social Inclusion show higher poverty rates for the elderly compared to the national average in the late 1990s in a majority of the EU member countries (Austria, Belgium, Denmark, Finland, Greece, Ireland, Portugal and the UK). The report also notes that households comprising single elderly persons run a higher risk of experiencing poverty than the national average, and that in several countries elderly women have Table 6 Elderly under the EU poverty line (%) ______Early 1990s Mid-1990s Czech Republic 1.4 4.5 Netherlands 5.8 11.2 Hungary 10.1 9.6 Denmark 12.2a — Germany 12.5a 8.6 Sweden 13.3a 4.6 Belgium 13.7a 15.4a Spain 13.7 — Finland 15.9a 6.2 Italy 16.9 14.8 Poland 17.3a 13.4 France lS.3a 14.7 UK 26.6a 18.1 Note: a Higher than national poverty rate. Source: LIS; own calculations. Much more starkly dian Figures 2 and 3, Figure 4 shows the predominance of other social transfers in reducing poverty. (The different heights of the bars indicate the importance of occupational and private pensions.) In a majority of the countries, the safety net contributed very little to absolute poverty reduction. Only in two countries - Denmark and the UK - does the importance of means-tested benefits stand out. The UK further stands out because of the large percentage of the elderly who were poor after occupational and state pension benefits. Typically during the mid-1990s these benefits reduced the poverty rate to between 5 percent and 15 percent of the elderly in the other countries, whereas the figure was 35 percent in Britain. The absolute measure of poverty reduction, however, conceals significant differences in the performance of the safety net in aiding the elderly. Looking at the relative poverty-reduction effectiveness scores, we find that the Danish score is slightly higher than the British one. In both countries around 50 percent of the pre-means-tested poor were lifted over the poverty line, in Finland and Sweden around 30 percent, in Hungary, Poland and Spain over 15 percent, and in the Czech Republic over 10 percent. The scores were around 5 percent or lower in the odier countries. To sum up, our analysis of the poverty reduction of vulnerable groups identifies serious holes in the safety net of several countries. In Italy, France and Spain the safety net offered little assistance to vulnerable groups. The analysis also points to shortcomings that are not immediately apparent when looking at the population as a whole. Both Belgium and Germany have poverty rates below the average and non-means-tested transfers account for the overwhelming portion of poverty reduction. However, when the regular social protection system failed (as in the case of the unemployed and the elderly in Belgium and solo mothers in Germany) the safety net was not much help to these groups. Similarly, Journal of European Social Policy 2002 12 (4) 320 Sainsbury and Morissens 100 so m a m D Other transfers S Safety net [1 Poverty rate CZ96 SW95 FI94 GE94 HU94 NL94 DK92 PL95 SP90 FR94 IT95 BE97 UK95 Country/year Figure 4 Poverty reduction for elderly (safety net and other transfers) Source: LIS; own calculations. an examination of the relative effectiveness of means-tested benefits reveals weaknesses in the British safety net, especially given the prominence of this type of benefit. Our analysis provides several additional insights. (First, in the European literature on poverty, there has been a tendency to distinguish between social insurance and social assistance in terms of their primary aimsyThe main goal of social insurance has been "the protection of the standard of living of persons exposed to social risks (such as unemployment, sickness and old age), whereas the major objective of social assistance is poverty alleviation. .Emphasis on this distinction is misguided because it obscures the significant role played by social transfers other than means-tested benefits in poverty reduction, as is evident through a comparison of Tables 1 and 2 along with the results reported in Figures 2, 3 and 4 (cf. Deleeck et al, 1992: ch. 5). In addition, the British case demonstrates the inadequacies of a poverty reduction stratég)' based primarily on means-tested benefits. Second, our analysis has focused on the poverty-reduction effectiveness of means-tested benefits, and we have used this measure as one dimension in determining how countries group together. This procedure reveals an important difference between countries representing the Social Democratic welfare regime and the Conservative Corporatist regime. Contrary to earlier assumptions and theorizing that means-tested benefits are of minor importance in the Social Democratic regime, our results indicate that this type of benefit pulled a larger share of the pre-means-tested poor out of poverty than in the other countries - even the UK. This difference would be difficult to detect using Esping-Andersen's regime typology because he has made means-tested benefits a defining property of one of his ideal types rather than a variation across welfare states. Elsewhere, however, he has argued that the role of the means test is limited in the Conservative welfare state model and minor in the Social Democratic welfare state model, suggesting the opposite of our findings (Esping-Andersen and Micklewright, 1991: 51). Journal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 321 Third, our analysis produces a cluster of strange bedfellows. Italy and the UK had the Jiigh^sxpOTejty_raies_(_EigureJ^)_and_theJowest— relative poverty-reduction scores (Table 1). In the mid-1990s the Italian and British social protection systems only lifted 4 out of 10 of the pre-transfer poor above the EU poverty line. However, as we have seen, this outcome resulted from fundamentally different policy constructions. In fact, their social protection systems are virtual opposites. Italy has yet to enact a national minimum income guarantee, while in the UK national social assistance programmes have assumed increasing importance as social insurance has been successively marginalized (Sainsbury, 1996: 205-12). Fourth, this difference points to two axes to classify or group the countries: (1) the effectiveness of means-tested benefits; (2) the effectiveness of other social transfers. This gives us four basic categories: low effectiveness of both the safety net and other transfers - Italy; low effectiveness of regular social transfers but high effectiveness of means-tested benefits -UK; high effectiveness of other social transfers but low effectiveness of means-tested benefits - Belgium; and high effectiveness of both -Finland. Towards a European safety net? The preceding analysis reveals a surprising amount of diversity in the basic safety nets across the European Union and their capacity to alleviate poverty. The inclusion of all the member and candidate countries would further magnify variations. This diversity suggests that the harmonization of safety nets of the member countries and future members may be as difficult an endeavour as attempts to harmonize social insurance schemes. The difficulties are compounded since the safety net is embedded in the larger framework of social protection in a country. However, a long-term trend during the past two decades is towards convergence in that the introduction of means-tested benefits has either complemented or replaced employment-related transfers. First, several Bismarckian —countries-with-rudimentary-safety-nets-have— embarked upon reform. As a response to the growing number of persons without social insurance benefits, France pioneered the RMI [revenue minimum d'insertion) in 1988 (Palier, 2000), which has since been the object of imitation. In 1991 Spain put in place new means-tested schemes - family allowances, old-age and disability pension improvements, and a benefit similar to the French RMI but administered at the regional level (Laparra and Aguilar, 1996: 97-106; Cousins, 1999: 129, 155-6). Italy also moved towards similar reforms in the late 1990s, introducing a version of RMI on an experimental basis and means-tested child allowances for large families (Ferrera, 2001: 183-4). Finally, Portugal, not included in this study, adopted a guaranteed minimum income scheme in the late 1990s. In effect, the idea of a safety net guaranteeing a minimum income has become a part of a 'European social model' emerging through the process of open coordination (Saraceno, 2001:5). Similarly, the post-communist countries had fairly rudimentary and often unofficial social assistance programmes prior to 1989. In the early transition period, the three candidate countries adopted new legislation, and spending on social assistance in absolute and relative terms climbed during the decade. More importantly, the Czech Republic and Hungary j replaced employment-related and universal'; child allowances respectively with income-tested ones, while Poland made child allow- j ances subject to means testing. Our analysis of ' the effectiveness of means-tested benefits in reducing poverty in the early and mid-1990s also indicates their growing prominence (Table 2). More recent data would no doubt \ show a further accentuation of this trend. Despite a general trend towards greater targeting and more means-tested benefits across countries, diversity in this area is simultaneously increasing. In coping with the widespread hardships resulting from the Journal of European Social Policy 2002 12 (4) 322 Sainsbury and Morissens transformation of their economies, the transition countries have come up with different rsafety=net-solntioTrs—whichroften-lack-a-coun^-terpart in member countries. Czech decisionmakers developed a system of income-tested state jjocial subsidies to guarantee minimum living standards that has focusgd on families with children. The Hungarians and Poles pursued an ad hoc strategy, relying on massive temporary emergency payments. The member countries also differ greatly in the extent to which their assistance programmes are general or categorical, and on this score divergence appears to be increasing. Recent reforms in many countries have tailored programmes to specific categories of need, while the Nordic countries continue to rely on general programmes and the Netherlands consolidated special assistance benefits into a single scheme in 1996. Moreover, deep-seated differences of opinion about means-tested benefits and safety nets exist. Perhaps the most fundamental political divisions revolve around the design of the resource test and the level of benefits. Should the test be an affluence or a poverty test? What should be the extent of family-support obligations in relation to the resource test? Equally controversial (if not more so) are benefit levels. Few countries have a safety net that provides benefits capable of lifting recipients over the EU poverty line. Neo-liberals and others equate generous benefits with work disincentives, and on these grounds they oppose raising minimum benefit levels. There is growing evidence, however, that a generous safety net does not necessarily lead to low employment rates. In their comparative study of solo mothers, Kilkey and Bradshaw (1999) found that Finnish assistance was among the closest to average earnings, but mothers had a very high employment rate. With respect to the pending enlargement of the European Union our analysis points in two directions. On the one hand, as the decade progressed, the candidate countries experienced rising poverty levels and poorer policy performance. On the other hand, the LIS data indicate that the three prospective -members-have-pover-ty-rates-simikr— to-those— of the member countries. The effectiveness of | their policies in reducing poverty is also in line \ with the EU countries - or even better. Furthermore, the three countries did not form a distinctive group that set them apart from the EU. Instead the Eastern European coun- . tries tended to cluster together with other i member countries. Acknowledgements We would like to thank David Jesuit who organized the panel 'Income Distribution and Redistribution in Cross-national Perspectives: New Insights from the Luxembourg Income Study' at the 2001 Annual Meeting of the American Political Science Association, which Jed us to write this article. In addition we thank Julia Lynch for her critical comments as discussant and subsequently. Notes 1 For an overview and evaluation of the European Poverty Programme, 1975-80, see Dennett et al. (1982). 2 We adopt a modified version of Esping-Andersen's regime typology, following critics who have argued that die Southern European countries represent a fourth regime type (Leibfried, 1992; Ferrera, 1996) and Van Kersbergen's (1995) scheme of including the Netherlands with the Continental countries rather than with the Nordic countries. 3 Subsequently Ian Gough (2001) has tested the robustness of the social assistance regime typology using cluster analysis. 4 LIS does include tax allowances in national tax estimates. In general, however, they are not identifiable, but they do have an impact on our results through lowering taxes. 5 Our analysis also differs from the model family income matrix used in the comparative study of social assistance in the OECD countries (Eardley et al., 1996; Gough et al, 1997) with regard to the treatment of costs of housing. In LIS, housing benefits are treated as income, and there is no information on housing costs for most of the countries (9 out of 13). As a result, we do not analyse disposable income after housing Journal nf Eurotiean Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 323 costs. However, in the British case we have checked the LIS data to see to what extent housing benefit covers the housingjosts-of-the-poor. A substantial share (roughly two-thirds) of housing costs is covered for those who receive the benefit (cf. DSS, 1994: 58). The problem is that only around 40% of the households below the poverty line received housing benefit (cf. DSS, 1994: 17, 48). In other words, it is very difficult to generalize about the disposable income of the poor after housing costs and the ■ replacement rate of benefits after housing costs in the UK. 6 For example, Barbara Hobson (1994) found that solo mothers in the United States and Germany had nearly identical income packages but their poverty rates diverged considerably in the mid-1980s. 7 Since our measures of poverty and poverty reduction are after taxes, they take into account any taxation of benefits. 8 Comparing the early and mid-1990s, three shifts occurred. France and Poland moved from the cluster of countries where means-tested benefits are least effective to the intermediate cluster in the mid-1990s. In the French case the RMI - an important new means-tested benefit - had only been introduced in the year prior to the first survey. In early 1995 Poland replaced employment-based family allowances with means-tested benefits; without dris reform Poland would be located in the cluster of minimal poverty reduction. The third change is that the Czech Republic moved to the third cluster, and this probably reflects the reforms adopted in 1995 when the new system of means-tested state social subsidies supplanted universalist policies (cf. Förster and Tóth, 2000). 9 There has also been much speculation about the negative effects of local administration, primarily the likelihood of differential levels of provision across municipalities. Interestingly, there is not much regional variation in the poverty rates in the Nordic countries. Regional differences were largest in Norway and smallest in Sweden (Gus-tafsson and Pedersen, 2000: 36, 93, 154, 189). 10 In a few instances solo mothers had higher poverty rates than the unemployed (Denmark 1992, the Netherlands 1992 and 1994, and Germany 1994), as did large families (Denmark 1992, Hungary 1994 and Germany 1994). With the exception of the Netherlands, the unemployed in these countries had relatively low poverty rates. References Abrahamson, P. (1997) 'Combating Poverty and Social Exclusion1, in W. Beck, L. van der Maesen and A. Walker (eds) The Social Quality of -------Eur-ope^-he--Hague:-lQuwerL;a^^lmetm&önäl. Atkinson, A. B. (1998) Poverty in Europe. Oxford: Blackwell. Bradshaw, J. and Terum, L. I. (1997) 'How Nordic is the Nordic Model? Social Assistance in a Comparative Perspective1, Scandinavian Journal of Social Welfare 6: 247-56. Bradshaw, J., Kennedy, S., Kilkey, M., Hutton, S., Corden, A., Eardley, T, Holmes, H. and Neale., J. (1996) Policy and the Employment of Lone Parents in 20 Countries. Universit)' of York: Social Policy Research Unit. COM/1999/347. 'Communication from the Commission: A Concerted Strategy for Modernising Social Protection' (July 1999). COM/2001/565. 'Draft Joint Report on Social Inclusion. Communication from the Commission to the Council, the European Parliament, the Economic and Social Committee and the Committee of the Regions'. Available at: http://europa.eu.int/comm/employmentsocial/soc-prot/soc-incl/indexen.htm (10 October 2001). Cousins, C. (1999) Society, Work and Welfare in Europe. Basingstoke: Macmillan. Deacon, B. (1992) 'The Future of Social Policy in Eastern Europe', in B. Deacon et al., The New Eastern Europe: Social Policy Past, Present and Future. London: Sage Publications. Deleeck, H., Van den Bosch, K. and De Lathouwer, L. (eds) (1992) Poverty and the Adequacy of Social Security in the EC. Aldershot: Avebury. Dennett, J., James, E., Room, G. and Watson, P. (1982) Europe against Poverty. The Europeaji Poverty Programme 1975-80. London: Bedford Square Press/NCVO. DSS (Department of Social Security) (1994) Social Security Statistics 1994. London: Her Majesty's Stationery Office. Eardley, T, Bradshaw, J., Ditch, J., Gough, I. and Whiteford, P. (1996) Social Assistance in OECD Countries, Vol. 1: 'Synthesis Report'. London: HMSO. Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism. Cambridge: Polity Press. Esping-Andersen, G. and Micldewright, J. (1991) 'Welfare State Models in OECD Countries', in G. A. Cornia and S. Sipos (eds) Children and the Transition to the Market Economy: Safety Nets and Social Policies hi Central and Eastern Europe. Aldershot: Avebury. Eurostat (1994) Poverty Statistics in the Late 1980s. Luxembourg: Office for Official Publications for the European Communities. Eurostat (2000) European Social Statistics, Income Poverty and Social Exclusion. Luxembourg: Office for Official Publications for the European Communities. journal of European Social Policy 2002 12 (4) 324 Sainsbury and Morissens Ferrera, M. (1996) 'The Southern Model of Welfare in Social Europe', Journal of European Social ___Lolicy-6J,l-):JJ-=i7-.------------------------------------------- Ferrera, M. (2001) 'Targeting Welfare in a "Soft" State: Italy's Winding Road to Selectivity', in N. Gilbert (ed.) Targeting^ Social Benefits. New Brunswick, NJ: Transaction Publishers. Förster, M. F. and Tóth, I. G. (2000) 'Trends in Child Poverty and Social Transfers in the Czech Republic, Hungary and Poland: Experiences from the Years after Transition', Luxembourg Income Study Working Paper No. 226. Syracuse, NY: Maxwell School of Citizenship and Public Affairs, Syracuse University. Gough, I. (1996) 'Social Assistance in Southern Europe', South European Society and Politics 1 (1): 1-23. Gough, I. (2001) 'Social Assistance Regimes: a Cluster Analysis', Journal of European Social Policy U (2)-165-70. Gough, I., Eardley, T, Bradshaw, J., Ditch, J. and Whiteford, P. (1997) 'Social Assistance in the OECD Countries', Journal of European Social Policy 7 (1): 17-43. Gustafsson, B. and Pedersen, P. J. (eds) (2000) Poverty and Loiv Income in the Nordic Countries. Aldershot: Ashgate. Hobson, B. (1994) 'Solo Mothers, Social Policy Regimes, and the Logics of Gender', in D. Sainsbury (ed.) Gendering Welfare States. London: Sage Publications. Kiikey, M. and Bradshaw, J. (1999) 'Lone Mothers, Economic Well-being, and Policies', in D. Sainsbury (ed.) Gender and Welfare State Regimes. Oxford: Oxford University Press. Laparra, M. and Aguilar, M. (1996) 'Social Exclusion and Minimum Income Programmes in Spain', South European Societ)' and Politics 1 (3): 87-114. Leibfried, S. (1992) 'Towards a European Welfare State? On Integrating Poverty Regimes into the European Community', in Z. Ferge and J. E. Koiberg (eds) Social Policy in a Changing Europe. Frankfurt: Campus Verlag. Ledemel, I. (1992) 'European Poverty Regimes', paper presented at the International Research Conference on Poverty and Distribution (Nov.), Oslo. Ledemel, I. (1997) The Welfare Paradox. Income Maintenance and Personal Social Services in Norway and Britain, 1946-1966. Oslo: Scandinavian University Press. Mitchell, D. (1991) Income Transfers in Ten Welfare States. Aldershot: Avebury. OECD (1998a) The Battle against Exclusion. Social Assistance in Australia, Finland, Sweden and the United Kingdom. Paris: OECD. OECD (1998b) The Battle against Exclusion. Social T™«,"/ „fV,,™t,omi W,V Vnlirv 900? 17 /4l Assistance in Belgium, the Czech Republic, the Netherlands, and Norway. Paris: OECD. JíalierjJB—(2aoa)-^iDefrosting^-the-Erench-JX?elfare-State', in M. Ferrera and M. Rhodes (eds) Recasting European Welfare States. London: Frank Cass. Rainwater, L., Rein, M. and Schwartz, J. (1986) Income Packaging in the Welfare State. Oxford: Clarendon Press. Sainsbury, D. (1996) Gender, Equality and Welfare States. Cambridge: Cambridge University Press. Saraceno, C. (ed.) (2001) Social Assistance Dynamics in Europe: National and Local Poverty Regimes. Bristol: The Policy Press. SZW (Dutch Ministry of Social Affairs and Employment) (1995) unemployment Benefits and Social Assistance in Seven European Countries. The Hague: SZW. Van Kersbergen, K. (1995) Social Capitalism: a Study of Christian Democracy and the Welfare State. London: Rontledge. Appendix This appendix describes the LIS data sets used in this article and offers more information about definitions of the vulnerable groups and the means-tested benefits variable. As can be observed from the Appendix Table 1, there were no changes in the type of survey used in each country, reducing intra-country comparability problems. Nonetheless, such problems exist, as is evident in very different values for the same variable in the two waves of data, e.g. the poverty rate of the unemployed in Germany 1989 and Germany 1994. One possible source of the problem is small sample sizes for certain categories in some of the surveys. Another difficulty is that LIS does not have information on gross wages and salaries for Hungary, France, Italy, Poland and Spain. Instead those datasets only provide a net wages and salaries variable. As a result, the poverty rates, based on market income, and the poverty reduction effectiveness are not fully comparable across countries and need to be interpreted with care. Poverty in Europe in the mid-1990s 325 Appendix Table 1 LIS surveys used for the analysis Belgium 1992 Belgium 1997 Czech Republic 1992 Czech Republic 1996 Denmark 1992 Finland 1991 Finland 1995 France 1989 France 1994 Germany 1989 Germany 1994 Hungary 1991 Hungary 1994 Italy 1991 Italy 1995 "' Netherlands 1991 Netherlands 1994 Poland 1992 Poland 1995 Spain 1990 Sweden 1992 Sweden 1995 United Kingdom 1991 United Kingdom 1995 _Socio-EconomÍG-Panel-of-Belgian-Households-Socio Economic Panel of Belgian Households Microcensus Microcensus Income Tax Survey Income Distribution Survey Income Distribution Survey Family Budget Survey Family Budget Survey German Socio-economic Panel Study German Socio-economic Panel Study Hungarian Household Panel Hungarian Household Panel The Bank of Italy Survey The Bank of Italy Survey Socio Economic Panel Socio Economic Panel Household Budget Survey Household Budget Survey Expenditure and Income Survey Income Distribution Survey Income Distribution Survey Family Expenditure Survey Family Expenditure Survey Definition of vulnerable groups Solo mothers are defined as households with a female head under 60 with at least one child aged under 18 present. Large families are defined as households where a father and mother are present, both 65 or younger, and with at least three children younger than 18. The term 'unemployed' refers to households where the head of household has a labour-force status that indicates that he or she is unemployed. For most countries we used LIS variable lfshd (labour force status head) and for some countries we had to use LIS variable dl8 (type of worker). The definition of 'unemployed' varies between countries. The 'elderly' are defined as those aged 65 and older within the population. Elderly couples are households where both the head and spouse are aged 65 and older. Single elderly women are defined as one-person households with a female head aged 65 and older. Single elderly men are defined as one- person households with a male head aged 65 and older. Means-tested benefits for the different countries Starting from the institutional information for countries in this study, we determined which programmes should be included in our means-tested variable. In a second step we looked for the availability of these programmes in LIS. Appendix Table 2 lists the programmes included in our means-tested variable, and they do not correspond to the LIS variable V25 (means-tested benefits). Other LIS variables have been used as well in order to get a more complete measurement of the role and effectiveness of means-tested benefits, in accordance with our definition of the safety net. Although we made a great effort to include all means-tested or income-tested programmes existing in the countries, problems related to data availability could not be 326 Sainsbury and Morissens Appendix Table 2 Means-tested benefits for the different countries Country Programmes included in means-tested benefits Belgium 1997 Czech Republic 1996 Denmark 1992 Finland 1994 France 1994 Germany 1994 Hungary 1994 Italy 1995 Netherlands 1994 Poland 1995 Spain 1990 Sweden 1995 United Kingdom 1995 Social assistance Supplement to social assistance Benefit to guarantee subsistence Guaranteed income for elderly Social assistance and other benefits Family allowances Housing benefit Social assistance cash benefits Rent subsidies Supplement for old-age benefit recipients Income support Unemployment benefits/basic amount and job market support Housing allowances for pensioners Social assistance Minimum guaranteed income Benefit for young child Allowance for single parents Rent subsidy Minimum old-age benefit Means-tested unemployment benefits Social assistance Educational stipends Housing allowances Regular and irregular social assistance Means-tested unemployment aid Means-tested child allowances/supplementary Economic support from central, regional, provincial, municipal bodies and includes maintenance, guaranteed minimum income, food allowance, etc. Benefits from National Assistance Act (ABW) Benefits for unemployed workers (RRW) Benefits for older and partially disabled workers (10AZ) Housing benefit Means-tested government transfers Family and care allowances Basic income Retirement and disability non-contributory pensions Social assistance Housing allowances for elderly and non-elderly Means-tested scholarships Income support Family credit Means-tested disability benefit Housing benefit Source: LIS survey and technical documentation. Tournal of European Social Policy 2002 12 (4) Poverty in Europe in the mid-1990s 327 entirely solved. Unfortunately LIS does not have all variables for all countries. In a few Instances aTTjenefit may be-available but not included in the survey and consequently it is not an LIS variable. In other cases it is not possible to separate a certain benefit because it is grouped with others. For Italy we were not able to isolate social pensions and means- tested child allowances and therefore they are not included in the analysis. "We were only able to use the regionally and locally administrated means-tested benefits, such as the guaranteed minimum income [Minimo Vitale), the food allowance {Minimo Alimentäre) and economic support {Assistenza Economica). Journal ofEurobean Social Poliev 2002 12/4) Welfare Regimes and ____lke-£xperienee--of------ Unemployment -in Europe Edited by DUNCAN GALLIE and SERGE PAUGAM €MFOMI> 1 The Experience of Unemployment in .Europe: The Debate Duncan Gallie and Serge Paiigam When the social status of individuals depends primarily on their participation in the systems of economic production and exchange of their society, there is a high probability that unemplovmeni-mlLLead to a loss of status and a feelingjDfjauiirA-e^r^^ v^öfö^pwcessoi what might be termed 'social disqualification' than a static state. It brings about a sharp drop in livingsTctndäTds, a weakening of social life, and marginalization with respect to those in work—effects which can become cumulative and lead to a situation of intense poverty and, at the extreme, of social rupture. It is notable that the\firstjmaior sociological enquiries on the unemployed date from the 1930gj[Laža"řsfeld et al. 1933; Bakke 1940a), at a time of economic crisis and unprecedented levels of unemployment. It was an issue that drew the attention of researchers much less in the period of economic prosperity after the war. Research focused much more on the effects of technical, social and cultural change on the social structures of Western societies. It was only towards the end of the 1970s that there was a renewal of research on unemployment by sociologists,1 economists and social-policy analysts. Our knowledge has expanded considerably over the 1980s and 1990s, but it remains highly fragmented since there was little co-ordination between the research carried out in different countries. Comparisons across countries in Europe are very rare, not least because they have had to confront the technical difficulty of the lack of standardization in the way data has been collected. In the absence of genuine-comparison,-the tendency has been to fall back on the mere juxtaposition of national monographs. This book seeks to advance our understanding by drawing on a research programme that has madean intensive effort to achieve a high level of comparability of data. The research, which lasted over aarfree-year period, hraught 1 In France the work of Dominique Schnapper enúUe&jĽépreuve.du chômage (1981) marked the beginning of a new phase in studies of unemployment; in Britain, the first major research programme on the social consequences of unemployment was carried out in theinid-1980s (see Gallie et al 1993). Duncan Gallie and Serge Paugam Table 1.1. Comparison of characteristics of unemployed (%) GER SW DK NL BE FR UK IRE IT GR S PT Average unemployment rates 1991-4 6.8 7.1 9.1 6.7 8.2 11.0 9.7 15.4 10.2 8.3 20.4 5.1 1995-7 9.0 9.4 6.5 6.1 9.6 12.2 8.0 11.3 12.0 10.4 22.0 7.1 Proportion of Iong-ternrunemployed 1991-4 37.2 8.5 29.8 47.4 58.3 37.1 38.2 60.0 61.6 49.9 48.7 37.9 1995-7 36.7 18.4 20.5 36.2 46.0 29.5 30.5 44.4 48.9 40.9 39.8 39.8 Average unemployment rates by sex 1991-4 Men 5.7 8.2 8.4 5.5 6.1 9.2 11.3 1.5.1 7.5 5.4 16.4 4.2 Women 8.3 5.9 10.0 8.4 11.3 13.3 7.7 15.8 14.9 13.3 27.4 6.3 Composition of the unemployed 1994 Males 49.5 59.8 48.2 53.8 47.0 48.0 66.5 62.1 49.1 42.7 51.5 48 0 Females 50.5 40.2 51.8 46.2 53.0 52.0 33.5 37.4 50.9 57.6 48.5 52.0 Age group 15-24 12.9 n.a. 22.2 27.7 25.2 25.0 28.3 31.2 39.5 39.0 33.1 34 2 25-49 60.6 n.a. 60.4 64.2 66.8 65.2 53.8 57.4 53.7 51.9 56.6 54.0 50+ 26.5 n.a. 17.3 8.2 8.1 9.9 17.9 11.4 6.8 9.1 10.2 11 8 Notes: GER: Germany, SW: Sweden, DK: Denmark, NL: Netherlands, BE: Belgium, FR: France, UK: United Kingdom, IRE: Ireland, IT: Italy, GR: Greece, S: Spain, PT: Portugal. Source: Figures for unemployment rates and long-term unemployment are drawn from European Commission, Employment in Europe, 1996, 1998. Figures on composition of the unemployed are from Eurostat, Labour Force Survey, 1994. n.a. = not available together research teams from eight countries: Denmark, France, Germany, The Netherlands, Ireland, Italy, Sweden, and the UK. It aimed to provide an overview of the experience of unemployment, partly byre-analysing national data sources with a view to increasing their comparability and partly by using the European Community Household Panel Study (ECHP),2 which started in-LS9$""and represents the first major source of information on the unemployed[m theEuropeanUnion.- The comparison of the experience of unemployment in Europe is all the. more interesting in that it addresses a wider concern by researchers to understand the forms of social regulation in European societies. To begin with, it is clear that, even though there are some commonTeatures, the relative risks of unemployment of specific categories of the population differs considerably between countries.- As can be seen in Table 1.1, women have a much higher risk of unemployment than men in several of the countries of Southern Europe—Italy, Greece, and Spain:—as well as in Belgium and France. While women are still disadvantaged, there is considerably less difference- in .the relative - For details, see Appendix Ä, pp. 37.Ď-80. The Debate 3 unemployment risks of men and women in Denmark, Germany, Portugal and Ireland. In contrast, in Sweden and the UK, it is men that are most affected. The effect of such sex differences in unemployment rates on the .composition ofLthejmemployed-depends in partonthe level of partieipa— tion of women in the labour market. At one extreme, in Greece, women represent 58 per cent of the unemployed; at the other, in the UK, only 34 per cent. There are also considerable differences in the age composition of the unemployed. Younger people,. aged 15 to 24, constitute a much larger proportion of the unemployed in the Southern European countries than in other countries. In Italy, Greece, Spain, and Portugal they represent more than 30 per cent of all unemployed people, compared with 22 per cent in Denmark and only 13 per cent in Germany. These differences point to the need to analyse the experience of unemployment not as something homogeneous, but as a phenomenon that takes place within particular economic, social and political structures and which, because of this, may have a different dynamic within each national culture.,. One factor that could well be important is the nature and the forms ofVl intervention of the welfare state. It is a plausible hypothesis that the living § standards of the unemployed wJlLderiend Jo a considerable extent on the systeni_ofjn2en2p]aymenŤ hengfits^JThe probability of people experiencing long spells of unemployment is also likely to depend on both the extent of development of active employment policies and on the availability of specific policies to assist women's employment, for instance through the provision of childcare. Further, one could hypothesize that the degree of social integration of the unemployed within society will depend on the form and the stability of family structures, in particular as these affect their capacity to provide opportunities for sociability and material or affective support. Finally, it is important to consider the experience of unemployment in the context of the pattern of economic development and the rapidity of sectoral change in particular societies, since the chance of finding work is likely to be heavily affected by the nature of demand in the market. The differences between European societies in terms of these three dimensions—welfare provision, the family and the market—are very great, and it would be worth recalling the most salient points of divergence. WELFARE REGIMES AND UNEMPLOYMENT The .conception of 'welfare regime' is taken in a broad sense. It refers to a system of public regulation that is concerned to assure the protection of inawiduals and to maintain social cohesion by intervening, through both legal measures and the distribution of resources, in the economic, domestic 4 Duncan Gallie and Serge Paugam and community spheres. The usefulness of any regime model for understanding the empirical pattern of welfare provision may differ between welfare domains, either because of the distinctiveness of the problems addressed or- because of Jhp- specific historical conditions at the time of institutional formation. It is quite possible, for instance, that the principles underlying the provision of welfare with respect to health may differ from those with respect to protection in the labour market. This wider issue of the degree of integration or segmentation of the principles underlying different aspects of welfare provision in a society is not one that can be pursued here. Rather it needs to be emphasized that we are concerned to use such types to explore one particular sphere of welfare provision. Our central concern is with those aspects of welfare regimes that provide protection from misfortunes in the labour market, primarily through the system of financial support for the unemployed and the institutional arrangements for intervention in the process of job allocation.3 The degree of coverage is likely to be a critical factor for the way the welfare state affects the experience of unemployment. It is, to a considerable degree, the very fact of receiving benefits to compensate for being without work that underlies conventional definitions of the unemployed. A person who is not receiving, or who has ceased to receive benefits is less likely to be socially recognized as a 'genuine' unemployed person. Coverage includes both those who receive insurance benefits and those who rely upon means-tested benefits. The balance between these can vary substantially between countries and, arguably, the nature of this balance may be important for the experience of unemployment. The higher the reliance on means-tested benefits, the greater may be the risk that unemployment will be stigmatic. In recent years, there has been a considerable expansion of means-tested assistance in most European countries, largely reflecting the deterioration of the labour market (Paugam 1999). But it should be noted that the link between these trends is stronger in countries where there is only weak coverage of the unemployed by insurance-based systems of social protection, revealing the importance of the underlying characteristics of the system. The level of financial compensation is also likely to have an important -effect on the experience of unemployment. In countries where a high level of replacement of earnings is provided over a relatively long period, the unemployed are more likely to be able to live in similar conditions to when they were in work. This also gives them greater opportunity to search thoroughly 1 This emphasis on domain specificity is one of the factors that distinguishes our approach from the influential analysis developed by Esping-Andersen (1990). It should also be noted that the concept of welfare Tegime used here relates to provision by public authorities, in contrast to Esping-Andersen's revised conception of a welfare- regime in terms of 'the. ways m which welfare production is allocated between state, market and households' (Esping-Andersen 1999; 73). The Debate 5 Table 1.2. Unemployment welfare regimes Regime Coverage Level & duration Active employment _________ _________________of coyer_______ ___ _„policy. 1. Sub-Protective Very incomplete Very weak Quasi non-existent 2. Liberal/Minimal Incomplete Weak Weak 3. Employment-centered Variable Unequal Extensive 4. Universalistic Comprehensive High Very extensive when they are looking for a job. It seems plausible that the degree of social stigmatization to which they are subjected will be lower, since there are fewer visible signs of their temporary loss of position. In countries where financial compensation is much more limited, the risk of poverty and of the cumulative growth of difficulties is likely to be much higher. Finally, the extent of development of active employment policies is likely to have an effect on the experience of unemployment, given that these can reduce the risk of long-term marginalizatipn from the labour market. When the unemployed have the possibility of improving their skills through training, they are likely to be in a better position to find a job. This argument cannot be accepted without qualification. When the number of available jobs in a region is limited, training courses can become little more than ways of keeping the unemployed occupied, giving them a minimum level of activity rather than providing any serious career perspectives. None the less, it has been shown that at least some types of employment policy which have been developing—albeit in different forms—across European societies in recent years can have real effects on job chances. In taking these three criteria—coverage, level of compensation and expenditure on active employment policies—it is possible to distinguish at least four "unemployment welfare regimes' in Europe: the sub-protective regime, the liberal/minimal regime, the employment-centred regime, and the universalistic regime. A sub-proteßtiyexeg.i.m.e is a system that offers the unemployed less than the minimum level of protection needed for subsistence. Few of the unemployed receive benefits, and when they do the amount is low. x\ctive employment policies are virtually non-existent. In this type of regime, it could be expected that the unemployed will experience severe financial difficulty and live under the poverty threshold. The probability of long-term: unemployment is also-'high, even thoiigh this is also likely to be conditioned by other factors such as the level and pattern of economic development. The liberal/minimal regime provides a higher level of protection for the unemployed than themb-prote^tive. It does not, however; cover those at risk of unemployment as a wholeand, above alL it provides a lov/ level of 6 Duncan Gallie and Serge Paugam financial compensation. Similarly, there is little development of active employment policies. This type of regime is also distinguished by its general philosophy. Whereas the sub-protective regime reveals a chronic absence of organized and-planned intervention by the state, the liberal/minimal regime rather reflects an explicit political will not to intervene too heavily in the protection of the. unemployed so as not to undermine the laws of the market. The underlying idea is to encourage the unemployed to take responsibility for themselves in order to avoid becoming dependent on social assistance. In order to minimize the possibility of a financial disincentive to work, there is a particularly strong reliance on benefits that are subject to means-testing, taking account of wider household income. It is a type of regime in which there is a strong risk that the unemployed will suffer from poverty. Given the emphasis on means-tested household income, it could be expected to provide particularly weak support for unemployed married women. The employment-centred regime provides a much higher level of protection for the unemployed than the liberal/minimal. The level of financial compensation is higher and the development of active employment policies reflects the concern of the public authorities to avoid the effective withdrawal from the labour market of the most disadvantaged sectors of the active population. Nevertheless, the coverage of the unemployed remains far from complete as a result of the principles of eligibility for compensation. These are primarily defined in terms of previous employment experience. The system is concentrated on those who have built up the greatest rights. It tends then to create a division between 'insiders' and 'outsiders'. Vulnerability to poverty is likely to vary greatly depending on the status of the unemployed person, as well as on the length of time they have been unemployed. It is a type of regime that will disadvantage people who have only been able to obtain temporary jobs or those with little work experience. It could be expected to disadvantage particularly sharply, women and younger people since these are less likely to have either lengthy or continuous employment experience. Finally, the universalistic regime is distinguished from the other three by the fact that it offers comprehensive coverage of the unemployed, a much higher level of financial compensation and a more ambitious active employment policy. Universalism also-tends to be associated with the individualization of rights. Benefits are granted relatively independently of the resources of other household-members. This should lead to both á low level of poverty among the unemployed and allower risk of labour market marginalization.:Given relatively jeasy eligibility rules and the individual basis of rights, this type of rejpme couldbe expected to discriminate the least between unemployed-people in terms of either sex or age: Welfare iegirnes nave been seen as ififferentiated: in terms of the extent to which they lead: to 'ďecQnrmoóxfícaiionV that is to say the progressive The Debate 7 detachment of the individual's status from the logic of the market (Esping-Andersen 1990). The introduction of modern social rights in capitalist societies has helped to give people resources that are independent from the market, thereby making them: other than-"merely an exchangeable commodity. But the extent to which this has occurred varies significantly between countries. With respect to the concept of 'decommodification', there is a clear gradient between the four types of unemployment welfare regime. The regime that is furthest from: achieving this is the sub-protective regime; that which is closest is the universalistic regime. A further common distinction is between 'beveridgian' and the 'bis-markian' systems, with the former characterized by the widest possible approach to social protection founded on the notion of social citizenship, and the latter by a more restrictive approach derived from participation in productive activity. It can be seen that the employment-centred regime reflects a bismarckian approach while the universalistic is closer to beveridgian principles. Both of the two other types of regime—the sub-protective and the liberal/minimal—are far from either principle. Finally, as will be discussed in more detail later, the regimes differ in terms of the extent to which they lead to de-familialization, that is to say ensuring the independence of the individual from reliance on other family members. This is likely to have particularly important implications for the experience of women (Lewis 1992; Orloff 1993; O'Connor 1996; Daly 1996). The universalistic regime stands out from the others in terms of the extent to which it provides support irrespective of other household resources. In considering this typology of unemployment welfare regimes, our view is that such models are best regarded as ideal-types rather than as descriptions of the institutional arrangements of particular countries. They are ways of describing the logic of particular processes of welfare regulation. It should not be expected that any country would represent a pure example of any such process. Rather, in reality, the welfare regimes of specific societies are likely to reflect, albeit to different degrees, a mixture of these different logics, and indeed their relative importance may change over time. However, there are grounds for thinking that the countries upon which we are focusing approximate more closely to one type of welfare model than another. We have taken three empirical indicators reflecting the different constitutive dimensions of these regimes. These^are: the proportion of the unemployed who receive benefits (see Table 1.3).rťhe average expenditure on benefits j>er unemployed person as a percentage of per capita gross domestic product (see Table 1.4), and finally expenditure on active^employment policies as a percentage of gross domestic product (see Table 1.5). These have been chosen as rjroviding a reasonable level of comparability and,, in the case of the last two, for giving a picture of the stability of the pattenr across time. They must however be regarded as providing approximate indicators. The issue 8 Duncan Gallie and Serge Paugam Taulľ 1.3. Proportion of unemployed in receipt of beließt Country Men Women Total Belgium 81.3 81.6 81.5 Denmark 66.9 66.2 66.5 France 47.6 42.8 45 0 Germany 75.4 65.7 70.5 Greece 10.6 7.1 8 6 Ireland 81.4 42.8 66 8 Italy 7.7 6.0 6.8 Netherlands 64.3 35.0 49 6 Portugal 29.2 25.5 27.3 Spain 32.3 15.6 23 8 Sweden 86.6 85.1 86.0 UK 71.8 36.6 59 4 Note: The unemployed are defined in terms of the ILO criteria. Source: European Labour Force Survey, 1995 (see Employment Gazette, 1995), except for Sweden. Figures for Sweden are drawn from register data for 1993 collected for the sample interviewed in the Level of Living Survey of 1991. Table 1.4. Expenditure on unemployment benefits per unemployed person (% of per capita GDP) Country 1980 1990 1993 Belgium 65.3 59.6 48.2 Denmark 88.0 61.3 61.8 France 38.6 33.2 36.1 Germany (West) 58.7 36.0 45.0 Germany* — — 46.8 Greece 64.9 32.6 31.1 Ireland 51.9 35.8 35.8 Italy 14.6 5.1 6.2 Netherlands 83.5 . 107.5 108.5 Portugal 13.2 " 9.4 23.7 Spain 79.6 52.5 73.7 Sweden n.a. n.a. n.a. UK 48.1 29.8 34.9 * includes the new Lander. n.a.: data not available. Note: Expenditure includes social assistance as well as insurance-benefits. Source: European Commission, 1995, Social Protection in Europe. of thelevel of compensation, in particular, has led to quite diverse estimates for particular countries of the extent to which unemployment benefit replaces income in work, depending on the: assumptions used. Tkeser-axe discussed further in Appendix 2. However; despite such variations, there: emerges a reasonably consistent pattern of the broad differences between countries. The Debate 9 Table 1.5. Expenditure on active employment policies (% of GDP) Country 1985 1990 Belgium 1.3 1.2 Denmark 1.2 1.3 France 0.7 0.8 Germany 0.8 1.0 Greece 0.2 0.4 Ireland 1.5 1.4 Italy n.a. n.a. Netherlands 1.3 1.2 Portugal 0.4 0.6 Spain 0.3 0.8 Sweden 2.2 1.6 UK 0.7 0.6 u.a.: data not available. Source: European Commission, 1998, Social Protection in Europe. It is clear, to begin with, that there is a strong similarity between the countries of Southern Europe.4 The proportion of the unemployed receiving benefits is lower than 10 per cent in Greece and Italy and lower than 30 per cent in Portugal and Spain.5 These figures are almost sufficient in themselves to justify considering these countries as closest to a sub-protective regime. If one takes, in addition, the criteria of expenditure per unemployed person as a percentage of per capita GDP, it can be seen that Italy is below the 10 per cent threshold, Portugal is lower than 25 per cent and Greece around 30 per centiOnly Spain stands out from this pattern at 73.7 per cent. Moreover, expenditure on active employment policies is also very low in these countries. It amounts to less than 1 per cent of GDP in Greece and Spain and to only 1.1 per cent in Portugal. Such policies are considered virtually Inon-existent at the national level in Italy (Saraceno 1992), which possibly explains why this indicator is not available in the national statistics. While the most notable feature of these systems is the gaps in their provision, it is possible none the less to detect a bismarckian influence. The Italian system of financial assistance for the unemployedJs interesting in this respect: it is characterized by a very strong dualization. It is a system that reflects.a particular way of handling-employment issues in very specific sectors of the economy, particularly hi large firms where employers and "* A. number of writers have- pointed to broad similarities in the structure of welfare arrangements in the:Southern.European countries (see.Liebfried 1992; Castles 1993" Ferrera 1996rSaracenol994), ^ The unemployed are defined in terms of the ILO criteria: people without employment, available to start work, and actively seeking employment. 10 Duncan Gallie and Serge Paugam employees have negotiated agreements that protect the workers and ai ihe same time reinforce their loyalty to the organization. Through the mechanism of the 'Cassa Integrazione Guadagnľ, employees of such firms receive 901 per cent: of their daily salary Tteotrg'h" the period-of-inactivity-Mo-reover, people covered by this scheme are not officially classified as unemployed. There are also special schemes for workers in industries characterized by a high level of seasonal variation in activity. With the 'Trattamento speciale di dissoccupazione\ employees from construction and agriculture receive an allowance representing more than half of their daily salary if they become unemployed—66 per cent in the construction industry, 60 per cent in agriculture—and this continues for a period of three, to six months, with possible further extensions. But, in Italy, as in the other Southern European countries, such protection has not been extended to the unemployed as a whole. The absence of more comprehensive coverage is likely to be particularly disadvantageous for young adults. The countries which are closest to a liberal/minimal regime are the UK and Ireland. The proportion of the unemployed covered by benefits is relatively high in the UK at 59 per cent, but expenditure on benefits is quite low: less than 30 per cent in 1990 and less than 35 per cent in 1995. It is also important to note the more and more restricted role of insurance benefits. The principal benefit 'Job Seekers' Allowance' is subject to means-testing from the seventh month, and the obligation has been imposed to subscribe to a 'Job Seekers Agreement'. All these measures have the objective, following the liberal logic, of encouraging the unemployed to actively seek work and to accept, if necessary, a precarious job. Since insurance benefit is for such a short period, the majority of the unemployed in the early 1990s were receiving means-tested benefit (Evans et cd. 1995). Ireland is also close to this type of regime, with relatively high coverage at 67 per cent, but low levels of ' benefit. It is, however, important to note that Ireland differed considerably f from the pattern in the UK with respect to expenditure on active employ- - \ ment policies. These represented 1.7 per cent of GDP in Ireland compared \ with only 0.4 per cent in the UK. ~ | Several countries approximate to the employment-centred model: France, j Germany, The Netherlands, and Belgium. The proportion of the unemployed | that receives benefits is high in Belgium (81.5 per cent) and in Germany (70.5 per cent) and lower in France (45 per cent) and The Netherlands (49.6 per cent). But the country order is no longer the same if one takes expenditure - ' on benefits. In this respect The Netherlands is highest at 108.5 per cent of _'-per capita income in 1993y followed by Belgium with 48 J2 per cent and Germany with 46.8 per cent. France is at an even lower level with 36.1 per cent. The case of The Netherlands consists then of two rather different t tendencies: a relatively low proportion of the unemployed- axe covered by insurance benefits, but these are quite generous for those who do get them. The Debate 11 This has the effect that large numbers of the unemployed have to look to social assistance for support. There are two such systems, one of which was created specifically for the unemployed, especially the long-term unemployed (Kemperman andVissers T999)rThe numbers on this have increased considerably in recent years, following closely the trend in the level of unemployment. The same process has occurred in France. A low proportion of insured unemployed-leads in a period of strongly rising unemployment to a sharp increase in the proportion receiving social assistance, especially the revenu minimum ď insertion introduced in 1988 (Paugam 1993). In Belgium and Germany where the coverage of unemployment insurance is more extensive, die increase in the numbers on social assistance has been less closely tied to changes in the labour market (Vranken; 1999; Breuer, 1999). Finally, it is notable that expenditure on active employment policies as a percentage of GDP is at a very similar level in these four countries and considerably higher than for countries closer to the first two types of regime. Such policies are an attempt by the public authorities to compensate for the gaps left in the system of insurance cover in a way that limits the pressure on social assistance schemes. Finally, the two countries that are closest to the universalistic regime model are Denmark and Sweden. The proportion of the unemployed covered by benefits and the level of benefits are relatively high in these countries. It should be noted that between 1994 and 1996, Sweden introduced measures to reduce the level of compensation, but it was subsequently increased again. These two countries have also been characterized for many years by a particularly strong emphasis on active employment policies with expenditures of 2.3 per cent of GDP in Denmark in 1993 and 3.2 per cent in Sweden. For instance, the 'activation' agreement signed in 1992 in Denmark included the objective of integrating young people rapidly into the labour market, by increasing the number of public sector jobs and offering training to all. It also aimed to develop incentives and opportunities for the long-term unemployed, through social development programmes, the creation of free zones, and the development of public employment. The overall package of these measures, in combination with the encouragement of early retirement, is thought to have made-a major contribution to reducing the level of long-term unemployment in the country. The discussion to date has been mainly concerned with the broad implications of the social protection systems for the unemployed taken as a whole. However the type of unemployment welfare regime may also have important implications for the degree of equality or inequality in the treatment of men and-women. Given that employment-centred systems: depend to a considerable degree on the direct contributions of employees, there is a risk: that women who have interrupted their careers for family reasons will be 12 Duncan Gallie and Serge Paugam disadvantaged compared to men with a more continuous employment record. Women could also be expected to be disadvantaged in liberal/ minimal systems by the reliance on a system of means-testing which takes account of overall household income. In contrast, this type of effect could be expected to be least strong in universalistic systems, due to their easier eligibility rules and their individualized system of rights. While it is difficult to find any simple metric for comparison, there are certainly important institutional differences between countries in eligibility rules (Rubery et al. 1998). The Danish system appears to be the particularly advantageous for women: only twelve months of contributions in the previous three years is sufficient to give eligibility for unemployment benefit, and this threshold has been reduced to thirty-four weeks for women who were in part-time work. It is notable that, although the unemployment system is a voluntary one, two-thirds of working women belong to it. In comparison, the British system is much less favourable for women. It gives greater importance to the amount than to the duration of contributions, which favours higher paid employees. As a result it takes a particularly long period for women to become eligible who are working in part-time jobs with low pay. Taking the empirical evidence on benefit coverage (Table 1.3), it is notable that the sex differential is very small in both of the countries closest to the universalistic model: Denmark and Sweden. It is also relatively low in three of the four countries that were seen as close to the sub-protective system—Italy, Greece, and Portugal. This, however, constitutes relative sex equality in the virtual absence of provision. In contrast, women were much less likely to receive benefit when unemployed in both of the liberal/minimal countries—Ireland and Britain. Indeed these are the countries in which the sex differential is greatest. Women are also less likely to receive benefit in the majority of the employment-centred societies. It should be noted, however, that there is considerable diversity in pattern: the sex differential is very sharp indeed in The Netherlands, but it is relatively moderate in France and absent in the figures for Belgium. While in general the evidence supports the view that such systems work to the disadvantage of women, there are clearly mediating factors that lead to important differences in the extent to which this is the case. Given the emphasis of such systems on employment experience, it is likely that one such factor will.be the pattern of women's labour market participation in specific societies. Overall,, these differences in benefit coverage for men and women confirm the earlier classification. Countries closest to the universalistic systems provide a high level of sex equality in coverage, while women are less well covered in the majority of countries closest to the employment-centred regime and in. both of those closest to the Ľbera^imrmTial regime. While this is the only aspect of welfare for the unemployed for which comparable evidence The Debate 13 for men and women is publicly available, it is clear that any full comparison of the degree of sex equality would require an assessment both of the level of benefits and of access to active employment policies. --Thereí are good reasons then --forLthinking-that the unemployment welfare regime is likely to have a major impact on the experience of unemployment. But itis also important to assess its.relativeimportance in comparison with the role that may be played by the pattern of family organization. THE ROLE OF THE FAMILY It seems likely that the nature of the family will have an effect on the experience of unemployment. For instance, it seems probable that an unemployed person who lives alone, one who lives with their parents or one who has responsibility for several children will not live the same experience. It may also be the case that, in a situation in which the male partner has a job, unemployment will have a different significance for a female partner in countries where the general level of activity of women is high than in one where it is low. In considering the possible impact of the family on unemployment experiences, the first factor that needs to be taken into account is the degree of stability ofThe familyas an institution. The trends with respect to the rates of marriage, divorce and births outside marriage, are generally used to provide a picture of the degree of 'de-institutionalization' of marriage and of the fragility of the traditional family model (Ŕoussel Í989). Earlier research has suggested that family instability may aggravate the precarity of work careers, especially for unemployed people. But it is striking to note how strongly these demographic indicators vary from one country to another, suggesting that the unemployed will experience very different family situations depending on the country in which they live (see Table 1.6). It is possible to distinguish at least three groups of countries with respect to the degree of de-institutionalization of the traditional family model. Sweden and Denmark share in common a high divorce rate and a high proportion of live births outside marriage: These are unquestionably the most advanced cases of de-institutionalization, even if the marriage rate is still high in Denmark.;The UK resembles these countries with its high divorce rate, but the proportion of "births outside marriage is somewhat lower. In contrast to die northern pattern, the Southern European countries, Greece, Portugal, Italy, and Spain, stand out with low'rates "both of divorce and of births outside marriage. Taking the 1980s cohort, the proportion of marriages ending:inrdiwarce was 12 per cent in Greece and only 7 per cent in Italy, compared with 46 percent in Sweden, Note that Ireland comes close 14 Duncan Gal lie and Serge Paugam Table 1.6. Demographie indicators Country Crude marriage Crude divorce Proportion of Proportion rate (1/1000) rate (171000) marriages ending in divorce for the 1980 cohort (1/100) of live births outside marriage (1/100) Belgium 5.1 3.5 34 15.0 Denmark 6.6 2.5 44 46.5 France 4.4 2.0 33 37.2 Germany Greece 5.3 6.1 2.1 1.1 33 12 16.1 3.0 Ireland11 4.3 (0.5) — 22.7 Italy Netherlands 4.9 5.3 0.5 2.2 7 31 8.1 15.5 Portugal 6.6 1.2 14 18.7 Spain Sweden 5.0 3.8 0.8 2.6 46 10.8 53.0 UK 5.5 2.9 42 33.6 a Given that divorce has been only recently legalized in Ireland, the figures presented are those of judicial separation rates (see Fahey and Lyons 1995). The legal separation rate for Italy tor the 1980 cohort has been estimated at 8.7%. Employment Precarity, Unemployment, and Social Exclusion (EPUSE) programme data. Source: Eurostat, data for 1995. to the southern pattern in. this respect. Even if the trends in most European countries have been similar over the 1980s and 1990s, the family model still remains more traditional in the Southern countries. Several countries occupy an intermediary position between the Scandinavian countries on the one hand and the Southern European countries on the other: this is the case for France, Germany, The Netherlands and Belgium. There are also major differences between countries in the roles attributed to the family. This is evident with respect to the responsibilities of the family for young children. These are likely to be heavily affected by the nature of welfare policies, which determine the .availability of childcare provision. The role of the welfare state in the construction of family models and their implications for inequalities between men and women has been emphasized by a number of writers in recent years (Lewis 1992; Orloff 1993; O'Connor 1996). Like unemployment benefit and employment/policies, family policies -differ considerably from one country to another. There is a need, then, to complement the emphasis on deeommodinca--tioEL, or detachment from the market, as a principle for comparing welfare regimes,, -with an emphasis on de-familialization^ or detachment from the family. Ä social policythat leads to de-familializatiop implies a commřtmeiit to collectivising the "weight of family responsibilities^ This is likely to be an The Debate 15 Table 1.7. Proportion of young children in publicly funded childcare (%) Country Aged 0 to 2 Aged 3 to school-age Belgium 20 95 Denmark 48 85 France -20 95 Germany 2 78 Italy 5 88 Netherlands 2 53 Sweden 32 79 UK ..2. 38 Source: Gornick et ai, 1997. important condition for the ability of women to give equal importance to their career and family lives. At the other extreme, it is possible to point to societies where the family remains both the pivot of sociability and the principal legitimate mechanism for taking care of the social needs of all of its members. It is striking to note, for example, the wide country variations in the socialization of children under school age in publicly funded childcare (see Table 1.7). This is a good indicator of de-familialization. The proportion of children under two years old with places in a creche is 48 per cent in Denmark and 32 per cent in Sweden compared with only 2 per cent in Germany, the UK and The Netherlands and 5 per cent in Italy. If one takes the proportion of children between 3 years old and school-age in publicly funded childcare, it is also weak in the UK (38 per cent) and in the Netherlands (53 per cent). It seems clear that women in the Scandinavian countries have much better opportunities to ensure the socialization of their children outside the family and thus to be less disadvantaged relative to men in their professional -careers. The nature of the family also differs very considerably with respect to the responsibilities for older children^and the length of time over which they continue to reside with their parents. When one compares the proportion of adult children of 20. to 29-years of age living with their parents, it is evident that the process of de-familialization is much more advanced in the Northern than in the Southern European countries or in Ireland (see Table 1.8). Over two-thirds of such young adults continue to live with their parents In Italy, Spain,-and Portugal, and more thahhalf in Greece and Ireland. The corresponding figure-is only 16 per cent in Denmark and 29 per cent in The Netherlands. If one takes-the proportions :amamj the unemployed, the divergences are even ^reater: :14 per cent in Denmark compared with 87 per cent in Italy. F, 16 Duncan Gallie and Serge Paiigam f Tahlľ 1.8. Proportion of adult children aged 20 to 29 living with their parents Country All 20 -29 year olds Unemployed 20-29 years olds Belgium 54.4 58.6 \~ Denmark 16.4 13.6 "~ France 39.5 48.5 - . Germany 35.9 28.6 '_■», Greece 57.4 78.1 -ŕ Ireland 61.8 71.6 if Italy 77.1 87.0 Netherlands 29.2 33.3 ¥ Portugal 69.8 74.2 ,X Spain 72.3 76.7 r UK. 32.4 41.7 í Source: European Community Household Panel (EC-HP), 1994, wave 1. | E. It is possible, then, to detect three models olfamily residence:, an extended ( dependence model, a model of relative autonomy between generations, and I finally a model of advanced inter-generational autonomy. t The extended dependence model characterizes a situation where different | generations are brought together in the same household under the wing of f the core generation. Adult children—and possibly older parents—are taken j care of within the same home. Sociologists and demographers have often § provided descriptions of this model emphasizing that it is based simultaneously on reciprocal exchange between the members of the household and on strict norms defining the obligations of each person within the group. It - -is a model that is reinforced by a very strong institutionalization of mar- . riage (Jurado Guerrero and Naldini 1997). While it restricts the autonomy of the individual, it has the advantage of offering everyone a minimum of protection. The long-term unemployed for instance would benefit from ■" direct family support. The relative autonomy model defines an intermediary situation between the extended dependence and the advanced autonomy models. In this ; model, adult children living with their parents are under an obligation,---- ; to look for an alternative solution. They must be actively preparing, their entry to the labour market. The normative system underlying this model is " • primarily one of autonomy, and individual responsibilities in the household/' are much. more, weakly defined. It is above all the difficulty that young;.-people encounter in-trying to enter the labour market that creates the neces- . , ; sity for parental solidarity. While this system is less rigid, it is also more _ ■ fragile. Adult children and older-parents live the experience as a temporary ; ~_ situation. The ties between parents and children can easily break up as a*" f result of misnnderstandirjgs in the household, especially when a relationship - The Debate 17 Table 1.9. Unemployment welfare regime and model of family residence, by country Model of Family Residence Unemployment Extended Relative Advanced Welfare Regime Dependence Inter-Generational Intergenerational Autonomy Autonomy Sub-protective Italy, Spain, Portugal, Greece Liberall Minimal Ireland UK Employment-centred France, Belgium Netherlands, Germany Universalistic Denmark, Sweden of dependence without possible exchange develops, which is commonly the situation for the young long-term unemployed. Finally, the model of advanced autonomy represents the normative system that contrasts most strongly with the extended dependence model. The self-realization of the young adult is regarded as inconceivable without acquiring autonomy from the parents. This does not of course mean that there is breakdown of affective ties. The children may remain to some degree financially dependent, but the norm is to live separately. This model allows people to cohabit with a partner earlier than is possible in the case of the two previous models. What is the nature of the link between the unemployment welfare regime i and the type of family residence model? We can begin by looking at the ( location of countries when the two dimensions are taken into account simultaneously (see Table 1.9). It is evident straight away that for the countries of Southern Europe there is a very strong correspondence between the sub-protective regime and the extended dependence model. It is also clear that there is a perfectly symmetrical relationship between the universalistic regime and the advanced autonomy family model in the Scandinavian countries. It might be tempting to conclude that there is a continuum between these two poles and to attribute the decisive role in the creation of family residence models to the type of unemployment welfare regime that prevails in each country. But any such conclusion would be over hasty. This is evident if one takes the example of the UK. Given"the minimal character of unemployment benefits and the liberal policies aimed ať keeping them low, it could be expected: that young people would: stay for a long time at home with their parents, much as is ihexase in freland. Sžmfariy;ľgiven its system of unemployment protection, it could be expected thai the situation 18 Duncan Gallie and Serge Paugam in The Netherlands and in Germany would be very similar to that in France and Belgium. But, in fact, The Netherlands and Germany, like the UK, are closer to Denmark and Sweden in their family residence models. It has to be concludedthen-thatthe dynamic of-family residence isnot entirely determined by the social protection regime, and that it is preferable to distinguish in our analysis the aspects of the experience of unemployment which are linked to the welfare regime from those which are linked to the family. For these reasons, we prefer to speak íil terms of welfare or unemployment welfare regimes when considering the sphere of intervention of the public authorities and of family models when examining the sphere of domestic life. When seeking to characterise the joint influence of several institutional factors linked to the experience of unemployment, we will use the notion of systems of social regulation. THE ROLE OF THE MARKET The concept of unemployment is inseparable from that of work, which provides the core interpretative framework in industrial societies. It is linked both to the idea of full employment and to the modern conception of paid work (Salais et al. 1986; Topalov 1994). Since unemployment is so closely linked to the conception of work that exists in an industrial society, its meaning may well vary depending on the idegree of industrialization and of economic development) In a region or a country with a relatively low level of economic development, the inhabitants may not even define themselves in terms of the employment relationship, in the sense that it has acquired in large-scale enterprises. They may consider their work as self-employed or carry it out, on an occasional or permanent basis, in the informal sector. Since work is primarily a matter of survival rather than a social status, the consciousness of being an unemployed person may be less strong, especially since formal financial compensation cm grounds of unemployment is negligible or even non-existent.6 Qualitative research in the less developed fi Pierre Baurdieu, in his research in Algeria in the early 1960s, found that it was only the peasants of the Kabylie who had internalized the conception of work common in industrial societies, largely due to the close links they had maintained with Western culture through their tradition of immigration to the large cities and to France (Böurdieu et-,ah 1963). Evea when they had been out at work in the fields for= long hours in the previous-days, they reported themselves as unemployed since they looked upon this agricultural work merely as a temporary way of getting by. It was only paid employment that they regarded as having true-value. The situation was completely different among, the .more traditional peasants of Southern Algeria. These considered work simply-in terms ofits^soeiälfunction and paid little attention to the notions of urietnplojmeiitar snb^einplwmeBtvevenJn periods when there was not a great deal of economic-activity and:they were only at work part of the: time. The Debate 19 Table 1.10. Economic restructuring in Europe. 1974-1993 (% of total employment) Agriculture Industry 1974 1993 Change 1974 31.2 1993 19.7 Change Belgium .3.8 2.5 -1.3 -11.5 ' Denmark 9.6 5.2 -4.4 21.4 22.3 0.9 France 10.6 5.1 -5.5 28.3 19.1 -9.2-, Germany 7.0 3.5 -3.5 35.8 28.5 -7.3 Greece 36.0 21.3 -14.7 18.5 15.6 -2.9 Ireland 22.8 12.7 -10.1 21.7 19.4 -2.3 Italy 17.5 8.0 -9.5 28.0 22.5 -5.5 Netherlands 5.7 3.9 -1.8 25.7 17.2 -8.5 Portugal 34.9 11.4 -23.5 ■ 24.9 23.8 -1.1 Spain 23.2 10.1 -13.1 26.3 23.0 -3.3 Sweden 6.7 3.4 -3.3 28.3 18.3 -10.0 . UK 2.8 2.0 -0.8 34.6 24.1 -10.5 1 Source: OECD (1997) Historical Statistics 1960-1995. regions of Southern Europe suggests that the conventional notion of unemployment is not as salient in the way people think about their society, in part because the 'unemployed' are mainly women or young people—see for instance Krief (1998). The adult men have the responsibility of ensuring the protection and survival of their families, and the economic system is organized to assist them to do this, even when conditions are harsh. It also seems likely that the experience of unemployment will be affected by the extensiveness and timing of sectoral restructuring of the economy. For instance, in all countries, the transition from an essentially rural to an industrial economy has been associated with major changes in the skills and work attitudes of the workforce. In much of Europe, this transformation was completed in the three decades following World War II, when there was an overall shortage of labour and there were possibilities for re-integration into othereconomic sectors. But much greater problems inevitably arose for countries for which severe reductions in the agricultural workforce coincided with wider economic crisis and an inability to generate jobs that could -compensate for those that had been lost. Unemployment became heavily concentrated on certain sectors of the population and involved severe marginal-ization from the labour market. Since the mid-1970s, the degree and nature of sectoral change has varied-very considerably between European countries (see Table 1.10). Talcing employment change in. the agricultural sector, it is notable that the Southern European countries Mve been by far the most affected in this period. The share of the agricultural sector went down from 35 to í 1.4- per cent m Portugal, and from 36 to 21.3 per cent in Greece. In comparison to a 20 Duncan Gallie and Serge Paugam f ŕ country such as France the restructuring of agriculture in Southern coun- i tries was taking place in a much more difficult overall economic climate. I The Mezzogiorno region in Italy represents a particularly extreme case | where structural decline in the agricultural sector occurred with little \ compensation in terms of the development of other types of stable economic í activity. I With respect to changes in manufacturing industry, it is clear that the | heaviest employment losses occurred in the more developed industrial [ societies. The share of such industries in overall employment went down | by about ten percentage pointsln Belgium, the UK, Sweden, France, and p The Netherlands. The reduction was less sharp in Germany, but it was still - [ considerable. Employment change in this sector was much lower in the | Southern countries: Portugal, Spain, Greece, and Italy. Finally, Denmark j". stands out as the only country in which there was a slight increase in the | share of this sector. } The European countries have then experienced different types and different ( intensities of shock to their employment systems in the period since the j mid-1970s. There are three broad groups of countries. The Southern coun- 1 tries experienced a particularly severe reduction of their agricultural sectors, I leaving endemic unemployment in certain regions. The societies that had developed large-scale industry earlier were subject to massive restructuring of the manufacturing sector. Unemployment in these cases was less likely to affect an entire sector of the workforce than to operate selectively, striking at those with lower skills and less ability to adapt to. technological change. Finally, countries such as Denmark saw much less radical types of economic rupture which would have made it easier to handle the problems thrown _ up by unemployment. The transitions between different types of work were , likely to require less drastic changes in skill and training, which should have helped to reduce the duration of unemployment spells. STRUCTURE OF THE BOOK Our starting point then is that the three types of factor that weTiave been -\ considering—the nature of social protection for the unemployed, the type of family system, and the pattern of economic change.—are each likely to ľ have major implications for the experience of unemployment. While they _,/ . are clearly not entirely autonomous^ they do not determine each other in a -,-'. way which can make it possible to simply logically deduce one from the other. -_ Rather, the specific configuration of the relations befeveen: them is likely to ';: ^ lead to significant differences in the experience of urieinrjioyjaaentrihetoeen^^ - The Debate 21 countries. In the course of this book, their impact will be examined with respect to three main dimensions of the experience of unemployment: financial deprivation, labour market marginalization, and social integration. The first part of the book focuses on the relationship between unemployffient and poverty. Chapter 2 looks at differences between countries in the risk of poverty for the unemployed and how these changed between the 1980s and 1990s. Chapter 3 examines the vulnerability to poverty and financial hardship of different types of unemployed people. Chapter 4 focuses on the situation of young unemployed adults, in particular considering whether family resources are effective in substituting for public welfare in countries where state provision is largely absent. Chapter 5 seeks to assess how far country differences in the risk of poverty for the unemployed can be attributed to the system of social transfers provided by the welfare state. The second part of the volume turns to the processes that may heighten or reduce the risk of labour market marginalization. Chapter 6 considers whether there is any evidence that the generosity of welfare benefits affects the attitudes to work of the unemployed, leading them either to attach a lower importance to having a job or to be particularly inflexible about the type of work they will accept. Chapter 7 assesses the impact of labour market conditions at the time of first entry to the labour market on later risks of unemployment, seeking to establish whether entry at a time of high unemployment leads to a long-term penalty in terms of job security. Chapter 8 focuses on whether there are variations between countries in the 'scarring effect', whereby people who have been unemployed have a higher risk of unemployment in the future. Chapter 9 considers the effectiveness of childcare policies in limiting the risk of labour market marginalization for the particularly vulnerable group of lone mothers. Chapter 10 considers how the role of social networks in determining the job chances of the unemployed may vary with different national labour market institutions. Chapter 11 involves a detailed comparison of Britain and Italy, to explore the influence of institutional differences in determining the type of people who are most at risk of long spells of unemployment. The third and final section is concerned with the issue of social integration. Chapter 12 looks at the impact of unemployment on people's social relationships in the household and the community. It considers how far unemployment is associated with social isolation and whether this varies between countries. Chapter 13 examines the tendency for unemployment and non-employment to become concentrated within households. Chapter 14 looks at the subjective side of social integration: the extent to which unemployment affects people's satisfaction with their lives. Chapter 15 seeks to assess how far the nature of the welfare regime affects differences between men's and women's experiences of unemployment. Finally, Chapter 16 looks at Individual approach in activation policy in the Czech Republic Tomáš Sirovátka 1 Introduction Demographic changes, globalisation processes, labour market transformations, individualisation and differentiation tendencies in society, economic limitations of the welfare state - all of these trends present new challenges to the welfare state. One of possible answers to these challenges is to set the objective of full employment, while requiring "activity" in the sense of participation in the world of work at the same time (cf. van Berkel and Moller 2002). The ;'Active Welfare State" concept (Giddens 1998) has become the guiding principle of welfare state's paradigmatic change, with "activation goals" permeating ever deeper through all its; domains. The change is explicit particularly in the European Employment Strategy - i.e. one of the key EU agendas - that can thus serve to verify the effectiveness of the "open method of co-ordination". For the present, the strategy of "activation" seems to be universally accepted - although with varying degrees of emphasis - in most EU countries, regardless then-go vemments' political orientation and the preferred welfare state regime. Generally speaking, the activation strategy seems to have support of the liberal, as well as the social democratic segment of the political spectrum - although its specific interpretations differ. The above-mentioned paradigmatic change within the welfare state is not as yet reflected in the post-communist countries where more converse tendencies can be seen. On the one hand, the social changes underway in these countries include "activation" of citizens and generate the need for activation strategies within the welfare state, on the other hand, the unfavourable economic conditions and inadequate capacity of public institutions often make implementation of necessary measures difficult. That is to say, the strategy of market transformation implicitly contains, among other things, citizen education towards individual respcnsibility, even by means of economic incentives. It also implies the need to cultivate citizens' human capital as a precondition of their future productivity. Besides, the This study was supported by the Grant Agency of the Czech Republic, grant No.: 403/03/1007 "Social Exclusion and Social Inclusion in the Czech Society". ** Schaol of Social Studies, Masaryk University, Brno, Gorkého 7, 60200, Czech Republic, phone: + 42 05 49496559, fax: + 42 05 49491920, e-mail: sirovatk@fss.muni.cz. 1 Drova their transformation strategy involves cuts in public spending (cf. Barr 1994). At the same time, transformation means new social risks that must be absorbed, which in turn necessitate^ new expenditures (cf. Offe 1996). Given the circumstances, risk-absorption efforts have so far centred primarily around redistributive and compensatory tools.1 On the contrary, "activation" measures - particularly active labour market policies - are lagging far behind the EU countries (cf. Gazes, Nešp 2003), primarily owing to insufficient resources and staff capacity necessary for effective implementation. However, compensatory strategies do not seem effective in the fight against new social risks. Public expenditures are on the increase, employment has declined and unemployment is high in the post-communist countries of central Europe. Such a state of affairs then forces political representations into taking unpopular steps: curtail social benefits, particularly unemployment benefits, tighten conditions of early retirement and proceed to other public finance reforms. The European union enlargement process involves implementation of guidelines laid down in the European Employment Strategy, thus drawing the accession countries' attention closer towards the issues of employment policies and activation strategies. The necessary motivation is generated by the European Commission's pressure in the form of the "open method of co-ordination" as well as by the countries' interest in gaining access to the European Social Fund. This is reflected both in the countries' National Employment Action Plans and in their Public Employment Services practice. Activation strategies in these countries are likely to assume a specific form depending on their respective ideological discourse and notion of activation strategy, as well as specific institutional environment and the process of implementation. What will activ strategies be like? In what ways will they be implemented and what will be their effect:?? To what extent will they be effective in including unemployed persons in the labour market and encouraging their social inclusion? This article aims at the evaluation of activation strategies applied in the Czech Republic's labour market. First, it outlines general approaches towards activation strategies and describes the role of individual approach in the process of activation. Then it explores the use of activation strategies in the Czech Republic and analyses their selected elements: 1 Instead, the growing unemployment was fought by pursuing the strategy of exclusion from the labour market (Offe 1985): encouraging early retirement and possibly prolonging the parental leave, along with introducing social assistance benefits and benefits that partially compensate for income depreciated by growing inflatipn and housing costs. their ation 2 specifically, the individual approach as it is represented by individual action plans implemented by employment offices. It analyses the extent to which the plans' implementation is influenced by the specific notion of activation strategy and by the circumstances of implementation. The analysis is based on case studies carried out in five local employment offices that run the individual action plans programme. 2 Activation strategies and the role of individual approach Activation strategies belong among central pillars of the welfare state reform that is now under way. Their prime aim is to integrate people fit for work in the system of paid employment in the highest possible numbers and throughout their whole life cycle.3 To this aim, a suitable combination of a variety of public policy, social policy and employment policy measures etc. is applied. Available literature on activation strategies distinguishes two model approaches to the goal of activation. These are a workfare approach versus insertion approach (Morel 1998, in Lodemel and Trickey 2001) or a workfare approach versus "Nordic"productivism (Esping-Andersen 1999) or a defensive versus offensive approach (Torfing 1999, in Lodemel and Trickey 2001) or a workfare model versus social inclusion model (Nicaise 2002) or a paternalist optimists' approach ("enforced participation") versus an activation optimists' approach - "inclusion through participation" (van Berkel and Moller 2002). Their basic features can be summarised as follows: For example, poor performance of public administration in the new member states is a subject of strong criticism by the European Commission. 3 Even though other forms of activation, such as voluntary work and education, are important as well (Berkel, Roche 2002). 3 CORE FEATURES WORKFARE MODEL SOCIAL INCLUSION MODEL Causes of unemployment and poverty Individual failure, poor work ethic Lack of skills Simple causes____________ Interplay of structural and culmral factors (labour market segmentation etc.) Multiple causes _________ Paradigm Rational Cultural, social-psychological rational Policy discourse Dependency, incentives, Welfare expenditure cuts, Obligations and individual responsibility___________ Social exclusion, social inclusion Social cohesion Ideology Right-wing Left-wing Policy approach „Orthodox consensus' Preventive „Reflexive activation" Curative Objectives of activation Labour market attachment Labour force flexibility Employability Social inclusion Target groups Long-term welfare state clients social welfare recipients, weakest groups_____________ Universal coverage + preferential treatment of the most disadvantaged___________ Principles of activation strategies Work first, enforcement, making work pay Balanced measures -income, training, work, empowerment________ Role of clients Objects of intervention Subjects/partners of interventipn Status of clients Subordinated, have duties, exposed to financial, administrative and legislative pressures_________________ Partnership, reciprocity, poses both rights and obligations Objects of activation Clients Clients and institutions The activation strategy can be characterised by further parameters related to the process of implementation: the degree of individualisation in the approach towards clients, the degree of differentiation of policies, the degree of decentralisation and the degree of pluralisation of objects of interventions. The two polar models (approaches) are characterised by differing degrees of the parameters above. Most importantly, if both approaches differ in the conception of objectives and principles of activation and in the definition of the role and status of clients, they also differ in the understanding of the importance, use and nature of individual approach towards clients. According to Nicaise (1995), approaches towards active labour market policy measures can be divided into two groups. Those focusing on multiple causes and disadvantages are a great deal individualised. On the contrary, those focusing on single causes and disadvantages (for example on insufficient professional skills) are usually standardised As concerns the first group, individualised and complex programmes that encourage motivation, self-confidence, knowledge, skills, experience, etc. are based on individual 4 approach grounded in precise identification of clients' needs and capacities, as well as in communication and active participation of the clients. Measures are tailored to clients' individual needs. "Activation through participation" (cf. Berkel and Moller 2002) represents a bottcm-up approach, where clients' rights are cherished and clients assume a role of active participants in the measures (subjects of activation). The measures proceed from clients' needs and capacities and count upon their participation in activation. This approach requires the choice of proper individual integration strategies, corresponding with individual combinations of disadvantages in the labour market and in other spheres. Individual approach playsi a central role both in setting individualised objectives of activation, and in suggesting and cipplying individualised measures. On the contrary, in the case of "enforced participation" based on a "top-down" approach, clients are perceived as objects of intervention and there is not much room for their activs involvement in activation - they are primarily expected to accept conditions of activation and submit to them. Measures depend above all on the capacity of institutions: this approach builds on the rational choice (based mainly on economic incentives) and works with a model of activation that can be applied regardless of the heterogeneousness of the unemployed and their individual needs. Individualisation is only applied in determining individual activation commitments and in inspecting whether they have been met by the clients. Such an individual approach is instrumental. APPROACH TO "INlJlVIDUALIZATION" WORKFARE MODEL SOCIAL INCLUSION MODEL Objectives of individual approach Prevention of dependency among specific categories of clients Social inclusion of concrete individuals Role of individual approach Formal, instrumental Substantial Choices for clients Very limited or non-existent Broad range of choices (forms of activation) Form of implementation Attention to clients' needs Stereotypical, formalised, limited individualisation Fully individualised, not necessarily formalised, centred around clients' needs Typical individualisation strategies Incentives (positive and negative sanctions) for categories of clients, Monitoring individual clients Universal, but individual: empowerment, human capital development The evaluation of the nature and implementation of individual approaches within activation strategies in the Czech Republic will analytically proceed from the above typology of 5 activation strategies and corresponding types of individual approach, without us expecting the typology to fully overlap reality. 3 Implementation of individual approach within activation strategies in the CR 3.1 The context and discourse surrounding the launching of activation strategies in ^;he Czech Republic and the role of the European union The nature of market transformation consists in re-commodification: in the privatisation of ownership, establishment of market exchange relations, including exchange of work for wage - and thus in activation of people in the labour market. No longer by means of "enforced employment" or obligation to work (applied under the command economy), but through individual interest and economic incentives. It can be interpreted as a labour market cash nexus where economic incentives motivate individuals to assume personal responsibility and engage in the labour market. It is therefore no wonder that from the very outset of the transformation great attention was devoted to the question of work incentives in the formal labour market in post-comqiunist countries. Besides, there was a range of other reasons: incentives are not very reliable new environment, given the low wages - especially in the secondary labour market -relatively well-accessible opportunities in the informal economy, failure of the public administration to effectively eliminate the grey economy and other factors - especially rooted "cultural" habits and reliance on the state inherited from communist timels are considered important. The growth in unemployment, particularly long-term unemployment,4 along with the booming grey economy led to a formulation of a hypothesis - commonly shared in the professional discourse of public services personnel - about a relatively significant proportion of "artificial unemployment".5 This hypothesis continues to be echoed in the mass med^a and shared by the public. In mid-1999 when unemployment amounted to 8-9%, 54 respondents in the Czech Republic agreed that unemployed people often or very misused unemployment benefits. Also, 54% stated that unemployed people often or 4 In 2003, long-term unemployment represented 40% of registered unemployment (and over a half cjf total unemployment according to the Labour Force Survey) in the Czech Republic. Common estimates by employment offices' personnel speak of a third of the unemployed. % Of often very 6 often had illegal jobs; 49% said that unemployed people were often or very often passive in searching for a job (Sirovátka 2002: 338).6 However, two groups must be distinguished within artificial unemployment: on the one hand it is unemployed people who are active in the grey economy, on the other hand it is "passive" unemployed who prefer to live on welfare benefits. Even though we lack enough facts and figures to support the hypothesis about the significance of the grey economy in the post-communist countries, everyday experience presents numerous arguments. Besides, there is indirect evidence: the Labour Force Survey data suggest a strong "entitlement effect" in that at the end of 2002 the number of registered unemployed people highly exceeded the number of people who said that they did not have paid employment and were registered in the labour force survey. The difference came to 38% of the unemployed registered with labour offices and was particularly marked in the category of short-term unemployed persons, especially men, which is indicative of possible engagement in the informal economy in parallel with the labour office registration. The hypothesis about "passivity" of the part of the unemployed is supported by the fact that the replacement ratio of the aggregate of available social benefits for low-income population groups in the CR is relatively high: for example, the replacement ratio for both two-parent and single-parent households with two children and only one employed family member whose earnings come to about two thirds of the average wage, is about 80% at the beginning of unemployment and 100% from the sixth month of unemployment. This is comparable for instance with the situation in Sweden (OECD 2002). Under the circumstances, the discussion about the need for activation centres around the issue of false unemployment. However, it shows that although artificial unemployment has been identified and named to be a problem by public employment services' and social depa-tments' personnel, staff shortages in employment offices and social assistance administration together with inadequate professional skills of the staff represent a major stumbling-block to its elimination. The Employment Act establishes that unemployed persons who reject a suitable job7 or fail to co-operate with the employment office may be struck off the register.8 Persons subject to such a penalty strike-off represented 10% of outflows from The Effects of Social Policy Transformation research, quota sample, N = 1319, School of Social Studies, Masaryk University. 7 A job with less qualification demands and a lower wage can be classified as a suitable job. However, unemployed persons' health and family situation (i.e. care commitments) are taken into consideration. 8 The law allows for a substantial decision-making leeway - e.g. upon breaking an appointment without excusing oneself, etc. 7 employment offices' registers in 2002. While the figure may appear high, it is far from reaching the estimated number of "passively" unemployed persons. The workload of 300-500 clients per a member of staff who is in direct contact with clients makes individual assistance in job-search, as well as inspection of job-search efforts, problematic, not to speak of lacking co-ordination of these institutions' operation. Given the increase in unemployment and in the number of welfare state clients, the relative staff capacity of these institutions has been on the decline. Under the circumstances, activation of clients was pursued through intensifying job-search inspections, tightening legislation that establishes entitlement to benefits and reducing the benefits replacement ratio.9 Besides, individual and group counselling has been advanced at employment offices, as far as the limited staff capacity of public employment services permits. All employment offices run Information and Counselling Centres that offer vocational and job-search counselling to young people. In addition, almost all offices haVe set up Job Clubs headed by highly erudite personnel (typically psychologists). Special activation programmes targeted at young people and based on a highly individualised approach are run at different parts of the country.10 Generally speaking, services for young people predominate in the sector, although Job Clubs for long-term unemployed adults exist as well. Given the limited staff capacity, working with young people appears to be easier and to have greater impact/Nonetheless, even these activities are still rather limited in scope - in comparison with unemployment figures. Between late 1996 and late 1999, registered unemployment increased from 34% to 9.4% and continued to grow further (to 10.3% at the end of 2003). Long-term unemployment increased from 20% to 40% of total registered unemployment (and to 50 % according to Labour force surveys) over this period. These figures rank the Czech Republic among countries with the highest proportion of long-term unemployment in Europe. The problem of benefit "dependency" became acute in 1999 since when 11-12% of households have been registered as repeated social assistance benefits recipients. In 2002, the average duration of drawing social assistance benefits reached 18 months (Sirovátka, Trbola 2003). In 1998, unemployment benefits were reduced from 60% to 50% and from 50% to 40% of previous wage (during the first three months and the following three months of unemployment, respectively). The subsistence minimum was valorised according to the growth in prices (with a delay), still lagging behind the growth in wages. Entitlement to unemployment benefits was restricted by imposing the requirement of a six-month period of uninterrupted employment for persons who had already been in receipt of benefits in the previous period. In 2003, the minister of labour and social affairs proposed to exclude school graduates from the entitlement to unemployment benefits. This proposal was rejected by the government but is going to be included into the new Employment Act in 2004. Nonetheless, activation efforts have so far centred largely around defensive strategies based on sanctions and on motivation through restricting entitlements to benefits. "Offensive" activation strategies grounded in the individual approach and building on active labour market policy programmes are being advanced, too, but are only applied in a rather limited scope. Counselling capacities of public employment services as well as active labour market policy fundi? are finite. In response to a threefold rise in unemployment, the Czech Republic did increase active labour market policy expenditures between 1997 and 1999, however, did not equal the standard common in the EU countries - the expenditures come to less than 0.2% of GDP, which is twice or even four times less than in countries with a comparable unemployment rate.11 Social insurance contributions allocated for the unemployment fund are three times higher than expenditures on unemployment, since the difference is used by the government to cover the deficient pension fund. Within the framework of the public finance reform in 2003, the government decided to permanently lower the social insurance allocation for labour market policies from 3.6% to 1.6% of the payroll, in favour of the pension fund, which certainly signals that the government does not intend to significantly increase expenditures on offensive activation strategies in the future. The approaching membership of the EU brings certain changes. On the one hand, the necessity to stabilise (that is restrict) public budgets in order to meet the entry criteria imposes real limitations on resources available for active policies. On the other hand, EU employment guidelines create pressures on the extension of active policies. This necessitates modifications of programme documents. In connection with preparations for the EU membership, the Czech Republic accepted a National Employment Action Plan in 1999 drafted according to the EU guidelines. In its first pillar (employability), the country took on a challenging commitment to meet the guideline 1 "prevention and activation" (later adopted by National Action Plans accepted in the following years). However, there was no interconnection between the National Action Plan and the national budget, the staff capacity of Public employment services did not change in any way, funds available for active labour market policy measures were not increased and counselling for unemployed people was not extended either. With increasing long-term unemployment and approaching accession to the EU the shortage of adequate individualised activation measures was gaining in weight. Therefore the 2002 National Action Plan openly stated that the Public Employment Service Administration It is for example a project Bridge, designed according to a British model, that involves individual diagnostics, counselling, employment training and subsidised employment with a follow-up career guidance. 11 In 2002, it was 0.17% of GDP, which is twice less than in Great Britain, three times less than in Hungary and four to six times less than in Spain, Finland or Germany. 9 did not have sufficient resources to meet the EES Guideline 1. Still, in the interest of meeting the Guideline 1 (prevention and activation) a decision was made to intensify counselling for the unemployed and extend employment services with the aim to reduce the growth of long-term unemployment. Considering the need for activation, this approach appeared to be a convenient and "inexpensive" instrument for eliminating passivity and benefits dependency seen among a part of the unemployed. Following the example of many European countries this intention was realised through Individual Action Plans. 3.2 Implementation of Individual Action Plans Individual action plans were piloted at 15 selected employment offices (out of 77 employment offices in the country) during 2002 and 2003. Five employment offices launched a programme "First Opportunity" for unemployed people under 25 years, five of them piloted a programme „New Start" for unemployed people over 25 years, and another five launched both programmes. The programme's implementation was subjected to continuous evaluation, the strategy was being perfected accordingly and a decision has already been made to launch IAPs at all employment offices and extend them universally to all unemployed people under 25 years, and later (beginning in 2006) also to the unemployed over 25 years. The following observations are based on case studies of IAPs' implementation at 1 3 selected employment offices. The conception of MPs' objectives When formulating concrete IAPs' objectives, employment offices accentuated above all better orientation and increased responsibility, independence and activity in job-search (by way of illustration, one of the employment offices suggested to title IAPs „Responsibility, Independence, Activity"). In addition, employment offices expressed the intention and set themselves the g