Children of Trust. Can trust explain the size and persistence of the welfare state? Carsten Jensen and Gert Tinggaard Svendsen 1 FIRST DRAFT

Transcription

1 Children of Trust Can trust explain the size and persistence of the welfare state? Carsten Jensen and Gert Tinggaard Svendsen 1 FIRST DRAFT Abstract Can trust explain the size and persistence of the modern welfare state? By analysing the cases of eight welfare states from the social democratic welfare regime (SDR) and the conservative welfare regime (CR), we suggest a model that links welfare institutions to the mode of trust. The model rests primarily on the work of Rothstein, Uslaner and Esping-Andersen and offers an explanation for an empirical puzzle, namely why the CR, which has relatively low scores in terms of generalised trust (GT), spends more money on public cash benefits and public old-age pensions than the universal SDR. The model argues that a positive feedback mechanism exists between the familiaristic welfare institutions and particularistic trust (PT) of the CR given that they fit together in a way that is likely to cause mutual self-reinforcement. This result suggests that future research should turn its attention to the institutional set-up of the welfare states and acknowledge that other types of social trust than GT can play an important role in this regard. Abbreviations: Social democratic welfare regime: SDR Conservative welfare regime: CR Generalised trust: GT Particularistic trust: PT 1 CJ: PhD fellow, Aarhus University. carstenj@ps.au.dk. GTS: Professor of Public Policy, Aarhus University and director of the Danish Social Capital Project (SoCap). gts@ps.au.dk

2 1. Introduction The welfare state is as much on the academic agenda as ever these years. Ever since the mid-1970s the fiscal pressure on the rich industrialised countries has been mounting and calls for the retrenchment or fundamental recasting of the generous welfare programmes of the Golden Age have been resounding. However, despite the strong factors pushing for such changes little has happened in terms of either retrenchment or path-breaking reforms; the institutions of the modern welfare states have proven remarkably resilient in most countries. It has been politically most troublesome to cut the size of the modern welfare state as voters simply want to maintain or even expand it (e.g. Pierson 1994; 1996; Castles 2004; Vis 2007; Paldam 2004; Nannestad, Svendsen and Svendsen 2007). According to Esping-Andersen (1990), the modern welfare states may be classified into three main types, namely the liberal, the conservative, and the social democratic model. First, in the liberal welfare model, the state plays a limited role providing the ultimate floor in cases where the market and civil society do not suffice. Furthermore, state-provided benefits are targeted, and concern about work incentives plays a dominant role. Second, the conservative model relies on the family and the employers as providers of social services. Private insurance schemes play a crucial role, and they are mostly tied to labour market participation. The activities of the state tend to be directed towards families rather than individuals. Third, the social democratic welfare model is characterized by universal welfare institutions. It has the state in a dominant role as the supplier of social services and benefits tend to be defined at the individual level but with differences depending on the individual s labour market history. The main financial sources are various forms of taxes (Andersen, 2004). Overall, the social democratic welfare model provides adequate benefits for (almost) the entire population such as pension system, child - 2 -

3 allowances, home help, student allowance, education, unemployment benefits, earlyretirement benefits etc. It is well known that the universal welfare model is highly popular (Larsen 2006). One explanation why this is so could be that it is perceived as being just by its citizens. This insight from Rothstein (1998) relates to the SDR in particular but which mutatis mutandis relates to the liberal welfare states as well. That is, in SDR it is perceived as fair that all, or most, citizens are eligible for rather generous benefits while we know that the liberal welfare states host the opposite perception (Larsen 2006). This way the universal welfare institutions in social democratic and liberal welfare states can be said to interact with, and to some extent even constitute, the moral foundations of the respective welfare states (cf. Uslaner 2002). Thus, the explanation of why the SDR persist even in the face of the mounting pressures to retrench seems to have a lot of leverage when comparing with the liberal welfare states. But can a similar argument be made in relation to the conservative ones? Given the central position of the Continental European welfare states we believe that it is vital that this question is not neglected, but rather is dealt with head on. Hence, the aim of the paper is to provide an explanation of how the trust-welfare institution relationship may work in the Continental welfare states as compared to the universal welfare states of Scandinavia. So far this has not been tried out systematically and the relationship between GT and the conservative welfare states are, thus, still left pretty much in the dark normally as some kind of (implicit) intermediary step between the liberal and the social democratic welfare states. Our contribution is therefore to focus on the comparison between the SDR and the CR and the role of trust in explaining their size and persistence

4 In relation to the CR, we add the trust concept of particularistic trust (PT). PT is, following Uslaner (2002), the trust in the nearest social group, most notably the family. We argue that this type of trust fits closely with the familiaristic welfare institutions of the conservative welfare states (cf. also Rothstein and Stolle 2003). The theory we provide is strongly inspired by the already existing work of the universal welfare institutions-gt argument, but also differs in important ways. At the most general level we argue that the emergence of a positive feedback mechanism is dependent on a close fit between the type of welfare institution and the type of trust in a given country. This formulation is in accordance with the existing argument of section 1, but allows for an extension that encompasses the conservative welfare states. We proceed by first suggesting two hypotheses in section 2. Section 3 then tests these hypotheses by operationalisation and measurement. Section 4 compresses the insights of theory and empirical evidence into a model and offers an explanation to why a surprisingly high level of welfare expenditure is observed for the CR. Finally, section 5 concludes the paper. 2. Existing hypotheses: outline and a preliminary test 2.1 Trust level As already touched upon in the introduction, it is the argument of the universal welfare institutions-gt literature that the universal welfare institutions and GT fit together. Such an approach is at odds with the traditional view where the link between welfare institutions and trust traditionally has been seen as a negative one. This so-called crowding-out hypothesis argues broadly speaking that large welfare state institutions are - 4 -

5 bad for GT because the latter has a tendency to substitute the former. Thus, a number of the societal functions that previously demanded, but also reproduced, social trust are now being undertaken by the welfare state; examples include help when unemployed and care in old age (Wolfe 1989; Taylor-Gooby 1998). Against this rather pessimistic line of reasoning a number of authors have pointed out that the continuing high levels of GT, and big welfare states in Scandinavia represent the disproval of the crowding-out hypothesis. According to these authors the relationship between GT and welfare states are actually the reverse of the one claimed by the pessimistic argument (quantitative evidence of the positive association is presented in Oorschot and Arts 2005). In short, universal welfare institutions and GT constitutes a positive feedback mechanism that mutually reinforces each other (Rothstein 2001; Rothstein and Stolle 2003; Kumlin and Rothstein 2005; Oorschot and Arts 2005; Käärlänen and Lehtonen 2006; Larsen 2007). Trying to account for the causal mechanism behind the universal welfare institutions-gt relationship, Rothstein (2003; see also Rothstein and Stolle 2003; Kumlin and Rothstein 2005) suggests that the more universal, uncorrupted and impartial the government institutions responsible for the implementation of laws and policies are, the more social capital you will get. It makes no sense to trust most people if they are generally known to bribe, threaten or in other ways corrupt the impartiality of government institutions in order to extract special favours (op.cit.: 69). In other words, wellfunctioning formal institutions reduce the likelihood that you will be cheated by facilitating the accumulation of trust. Furthermore, Rothstein (2003) suggests that the equal access to public goods in the Scandinavian welfare state also increases trust by reducing social cleavages. Partly in accordance with this line of thought, Larsen (2007) argues that in universal welfare states, the larger part of the population is eligible for welfare benefits

6 Contrary to this, in liberal welfare states only a small minority of the population are eligible, which entails that a clear divide emerges between those receiving and those paying for welfare. This is in turn argued to lead to a greater level of distrust in the wellfunctioning of welfare institutions as the majority will start questioning the motives of the recipients. While both institutional quality and equal access to public goods indicate an institutions matter approach where causality runs from institutional setup to the level of social capital we would, however, argue that the direction of causality could run the other way as well. Admittedly, institutional setups are crucial and allow countries to benefit from high stocks of social capital that facilitate and maintain extensive welfare states. Such hightrust countries gain a competitive advantage by saving resources when a substantial part of the population does not need much monitoring as they are likely not to free ride and behave in an honest manner. Overall, the transaction costs of supplying public welfare goods are lowered. As such, Nordic welfare systems are not shot down due to the potential free rider costs that would be devastating to the system in less trusting countries. Elsewhere, Kumlin and Rothstein (2005) direct a similar attack on the civic society approach in Putnam (1993), turning the institutions matter argument into the policy recommendation that, when investing in social capital, governments should increase the quality of political institutions rather than supporting voluntary civic associations (op.cit.: 362). This idea is confirmed by the findings in Nannestad and Svendsen (2007) where the quality of institutions are shown to matter for trust levels among immigrants in Denmark. One of the greatest achievements of this positive feedback argument of universal welfare institutions and GT is that it seems to fit very nicely with the often observed difference between the social democratic welfare states of Scandinavia and the liberal welfare states of the Anglo-Saxon world. That is, in Scandinavia very high levels of GT - 6 -

7 correlates closely with high levels of welfare effort, while the opposite is true in the liberal welfare states. Given the lack of universal welfare institutions in the liberal welfare states, a positive relationship, thus, seems to be established between the type of welfare institution and degree of GT. Given the lack of universal welfare institutions in the CR we should therefore also expect that this type of welfare state has a lower degree of GT than the SDR. Hence, our first hypothesis is that: H1: SDR countries have higher GT than the CR countries. 2.2 Size of welfare effort The second hypothesis follows naturally from the first, at least given the positive feedback argument of universal welfare institutions and GT. In countries where the level of GT is very high it is easier to up-hold large welfare states because these are perceived as being fair by the general public. As discussed in the first section this perception may be due to at least two different mechanisms (dealignment of the middle class from the poor, or lack of public administration impartiality), but the fundamental point is that public support dependent on how fair the welfare institutions are believed to be. Hence, in liberal welfare states, especially the USA, the welfare institutions are perceived as unfair by the large majority, which has eroded the political basis for maintaining pervious levels of welfare effort. Conversely, in the SDR the high levels of GT have been instrumental in maintaining the political basis for an extended welfare effort. Thus, given H1 the second hypothesis is that: - 7 -

8 H2: SDR will exhibit a much higher level of welfare effort than the CR 2.3. A preliminary test We will now present a preliminary empirical test in order to test the two hypotheses derived in Section 2.1 and 2.2 above. The countries we include in the analysis are core members of the conservative (Austria, Belgium, France, Germany, and Italy) and social democratic (Denmark, Sweden, and Finland) welfare regimes respectively. Norway is excluded due to missing variables. TABLE 1 ABOUT HERE Concerning the first hypothesis, Table 1 clearly documents that SDR do indeed have higher GT than the CR. The CR are characterised by a relatively low level average of GT (29.1%) compared to the average in the SDR (61.5% GT). France ranks strikingly low, whereas Denmark and Sweden top the list. The second hypothesis was that CR would exhibit a much lower level of welfare effort than the SDR when the latter hosts much more GT than the former (the confirmed H1 above just documented that). Alas, as it is, this is not the case as Table 2 shows by presenting various measures of welfare effort from the eight countries. In fact, when it comes to over-all public expenditure on cash benefits and public old-age pensions there is even a negative relationship in the sense that CD spends more than SDR. Thus, hypothesis 2 is rejected when considering the five classic conservative welfare states of Austria, Belgium, France, Germany, and Italy. Table 2 shows the country-specific data

9 TABLE 2 ABOUT HERE We face an empirical puzzle: If the argument of a positive feedback relationship between GT and universal welfare institutions is correct one ought not to see this. What is more, as has been documented in a great number of more qualitative oriented studies from the welfare state literature, the conservative welfare states are very change resistant; perhaps even more so than the social democratic welfare states (Esping-Andersen 1996; Pierson 2001). But how can this be if the lack of GT must lead us to expect that the public in these countries does not perceive large welfare spending as the right thing to do? In the next section we try to outline an answer to this puzzle. 3. Model The explanation of why social democratic welfare states persist even in the face of the mounting pressures to retrench seems to have a lot of leverage when comparing with the liberal welfare states, but runs into problems when compared with the conservative ones. As Table 1 and 2 in combination documented, generalised trust has no positive explanatory power at all when we focus exclusively on differences between SDR and CR countries in terms of welfare effort; in fact, total public cash benefits and public old-age pensions is negatively correlated with generalised trust. The theory we provide to account for this apparent anomaly is strongly inspired by the already existing work of the universal welfare institutions-gt argument, but also differs in important ways. At the most general level we argue that the emergence of a positive feedback mechanism is dependent on a close fit between the type of welfare - 9 -

10 institution and the type of trust in a given country. This formulation is in accordance with the existing argument of section 2, but allows for an extension that encompasses the CR. Our argument is deceptively simple; in conservative welfare states welfare effort is designed in a way quite distinct from the design of the SDR. Contrary to the universalism of the latter, CR have designed institutions that are aimed at maintaining the traditional family, i.e. the welfare institutions are familiaristic (van Kersbergen 1995; Esping- Andersen 1999). The familiaristic welfare institutions are characterised in particular by is emphasis on the male-breadwinner and the care duties of the domestic sphere. Hence, social protection if often biased against female labour force participation because it is made dependent on being member of the high status core workforce. The penetration of women into the core of the workforce, however, is obstructed by a number of interrelated factors. First and foremost, given the emphasis on domestic care duties the social care sector, i.e. public child and elder care, has seen comparably little expansion. This not only prohibits the removal of care duties out of the family, but also restrains the job growth in the sector most hospitable to expanding female labour force participation, the social care sector. These policies are further strengthened via benefit and tax structures that generally favours the traditional family, e.g. via providing special tax breaks for single-earner households. All in all, such familiaristic welfare institutions stand in sharp contrast to the universal ones that have been more or less defamiliarised, i.e. where benefits have been made independent of the family. The familiaristic welfare institutions, moreover, are also in strong discordance with the notion of GT. GT concerns the openness towards and trustworthiness of the generalised other whereas the familiaristic welfare institutions, as the almost exact mirror image, aims at protecting the closest social realm against the other

11 We propose to solve the puzzle by applying the concept of particularistic trust (PT) as an alternative to GT. PT is, following Uslaner (2002), the trust in the nearest social group, most notably the family. We argue that this type of trust fits closely with the familiaristic welfare institutions of the CR (cf. also Rothstein and Stolle 2003). Thus, conservative welfare states aims at up-holding the traditional family by maintaining familiaristic welfare policies, which is argued to be in accordance with the dominant mode of trust in these countries, namely PT. Given the close match between mode of trust and welfare institutions we should expect that the CR, like the SDR ones, are perceived as just or fair by a majority of their populations. This ought in turn to give rise to a positive feedback mechanism helping to up-hold both the welfare institutions and the mode of trust. Figure 1 shows a general model of the creation and persistence of welfare states also offering an explanation to the empirical puzzle identified when testing hypothesis 2 above. FIGURE 1 ABOUT HERE The first two boxes stem from Uslaner (2002; 2004). He argues that the mode of trust (GT or PT) is determined by the culture/cultural basis of a country rather than on government policies and institutional setups. He argues that GT is learned early in life from parents and school and that it is largely stable throughout one s life. In fact, parental trust can be traced back to grandparents trust, further and further back, suggesting deep cultural roots. Therefore, immigrants can carry social trust from their home country to their new homes and stick to their cultural traditions of trust rather than simply adapting to their new habitat. GT is not only carried over from one generation to another but also within ethnic groups moving from one nation to another (ibid.). Uslaner moves on to predict, based on

12 the GT measures from the World Value Surveys, that the most trusting people in the United States would be of Nordic background, regardless of where they live. Nordic ancestry does lead to greater trust. If your heritage is Swedish, Norwegian, Danish, or Finnish, you will be almost 10 percent more likely to believe that most people can be trusted (ibid.). Thus, it seems that where your ancestors came from matters more for trust than who your neighbors are now; how recently immigrants came to the United States does not seem to matter. British heritage makes you almost 5 percent more likely to trust others, confirming the old saying: An Englishman s word is his bond. In contrast, there are no significant effects for people with Eastern European, Russian, French, Italian, or even German ethnicity. In Continental Europe the culture emphasises the family and nearest social group, while in Scandinavia we traditionally emphasise larger groups. Uslaner traces these cultural roots to religion as a main determinant. Societies that are predominantly Protestant, namely the Nordic countries, are 28% more trusting than the least Protestant countries in Southern Europe Uslaner (2002: 232). Other authors (Bjørnskov 2005; Knack 2003; Svendsen and Svendsen 2003) have also indicated that GT is a stable cultural feature of society as well that only changes slowly over time. They also point to religion as one main determinant for GT. Such general idea that the cultural characteristics of a society will influence the economic development can be traced back to Weber (1930 [1906]) when he presented his famous theory of the Protestant ethic and the spirit of capitalism. Obviously, this does not explain why social capital generating institutions actually work in Scandinavia but not necessarily, for example, in Brazil or Bulgaria; or in third world countries in Africa and Southern America, which have received considerable amounts of development aid since World War II, including organizational training

13 Therefore, as Putnam (1993) says, we may assume that civic traditions are accumulated (and lost) through long, historical processes. This may exactly be the explanation, why the Nordic countries enjoy high levels of trust which, to some degree, have insulated these nations from non-cooperative behavior and free-riding. The last two boxes we got from Esping-Andersen (1990; 1999). As has already been discussed, Esping-Andersen argues that the type of welfare institution in a country (universalism or familiaristic) is determined by what political ideology is dominant in a country (see also van Kersbergen 1995). Thus social democratism leads to universal welfare institutions and conservatism leads to familiaristic welfare institutions. The positive feedback-mechanism from welfare institutions to mode of trust is from Rothstein (2001) as presented in section 2; i.e. after the creation of the welfare institutions a positive feedback mechanism sets in the way we described above. Our contribution is then to combine these three different theories. The argument is that an ideology can only become dominant in a country if it fits with the cultural basis and thereby also mode of trust in that specific country. This idea is similar to Weber s classic argument that the ruler must be legitimate in order to stay in power (Weber 1930). To exemplify, conservatism has not become the dominant political ideology in Scandinavia because it does not fit the culture and dominant mode of trust. 4. An empirical test In section 2 we presented Table 1 and 2 in order to show that the CR constituted a special case. In this section we will provide a more elaborate empirical test of our theory. We will use the eight countries already introduced and will move from a descriptive analysis to a data generated one. Since we believe that two different set of causal processes have caused

14 and up-holds the two welfare regimes we choose to rely on iterative partitioning cluster analysis. This technique has the nice property of not assuming a one-dimensional causality in the data as do regression analysis (Shalev forthcoming) and is therefore very suitable for analyses of welfare regimes (e.g. Obinger and Wagshal 2001; Powell and Barringtoes 2004; Jensen 2007). When applying iterative partitioning cluster analysis the countries are clustered into a predetermined number given the underlining dimensionality of the variables. Hence, we choose to form two clusters hoping that (a) the countries of each cluster corresponds to the two welfare regimes, (b) that the CR cluster exhibit a high level of particularistic trust and a low level of generalised trust, while the other cluster performs opposite, (c) that both clusters are big in terms of welfare effort, and (d) that the CR cluster is characterised by markedly more familiaristic welfare institutions than the SDR one. Relating to the last expectation, i.e. that the welfare effort to have a familiaristic tilt: Such qualitative-oriented features are generally quite hard to capture, but one measure that ought to do the trick is the level of welfare effort on social care services. As argued above, social care services have the double quality of both providing women with a job market as well as an opportunity to shift the care duties of the domestic sphere to the public. In this sense extensive social care services almost stand as the essence of defamiliaristic welfare institutions (Jensen 2007). We therefore expect conservative welfare states to be characterised by a low level of welfare effort on social care services compared to the social democratic welfare states. In order to measure the level of generalised and particularistic trust in the eight countries we rely on data from the European Values Study (the survey) and tries to replicate the operationalisations suggested by Uslaner (2002). We use the standard measure of generalised trust, i.e. the percentage stating that most people in general can be

15 trusted, while relying upon Oorschot and Arts (2006) composite index of family trust as our measure of particularistic trust. The index captures the degree to which individuals emphasise family relations and can take the values The measures of welfare effort are sampled from OECD s Social Expenditure Database and Tax-Benefit Models. 2 Table 3 summaries the results providing the means (so-called k-means) for the two clusters on the six variables. Table 3 ABOUT HERE The results are clearly supportive of the expectations. First of all, the countries group together in the SDR and CR respectively. Second, the CR is indeed characterised by a low level of generalised trust and a high level of particularistic trust compared to the SDR Third, in terms of general welfare effort the CR countries are either bigger than or as big as the SDR countries. Fourth and finally, in terms of social care services specifically then CR countries are clearly outperformed by the SDR countries. 5. Conclusion Can trust explain the size and persistence of the modern welfare state? Our argument builds on the observation emphasised by the existing literature that there exists a close match between the universal and impersonal institutions and the generalised, impersonal trust (GT). We first argued that the universal welfare institution and trust argument constituted a valuable corrective to the pessimistic crowding-out hypothesis as well as an interesting explanation of why the social democratic welfare states of Scandinavia have been so

16 persistent in the face of mounting challenges; due to welfare institutions in these countries that emphasise universalism a positive feedback mechanism has emerged between this type of welfare state and generalised trust. This is taken to explain why social democratic welfare states are able to maintain very large welfare states, while liberal welfare states, especially the United States, have been retrenching somewhat. Looking at the CR we argue that the structure of the welfare institutions are very different; instead of stressing universal features these institutions stress much more familiaristic. That is, the welfare institutions of Continental Europe mirror the conservative ideology that has been dominant in these countries and which emphasises the traditional role of the family. These familiaristic welfare institutions in turn seems to match closely with a certain mode of social trust often overlooked, namely particularistic trust (PT), i.e. trust that extends to the family and closest social group only. We showed empirically that the CR are indeed characterised by (a) high levels of PT, but low levels of GT (i.e. the polar opposite of the SDR), (b) bigger welfare states in general, and (c) more familiaristic oriented welfare state policies in particular than SDR. As demonstrated, GT has no positive explanatory power at all when we focus exclusively on differences between SDR and CR in terms of welfare effort; in fact, total public cash benefits and public old-age pensions is negatively correlated with GT. In sum, we have provided some preliminary evidence that PT covariates with familiaristic welfare institutions the same way GT covariates with universal welfare institutions. Both the SDR and CR are characterised by two different types of positive feedback mechanisms both of which can be part in the up-holding of big welfare states because they both generate support from the public via the perception of the fairness of the institutional set-up

17 However, the quality of the data does deny us the possibility of testing the propositions more dynamically, i.e. over time. Most importantly, we are not able to test directly whether there actually is a positive feedback relationship between the different modes of trust and the different welfare institutions. Yet, given our model, which primarily rests on the work of Rothstein, Uslaner and Esping-Andersen, this seems the most plausible account of the empirical results. In other words, future research should turn its attention to the institutional set-up of the welfare state and try to disentangle how it interacts with different types of trust

18 References Bjørnskov, C. (2005) Investigations in the Economics of Social Capital. PhD thesis, Aarhus School of Business. Castles, Francis G. (2004). The Future of the Welfare State. Oxford: Oxford University Press. Esping-Andersen, Gøsta (1990). The Three Worlds of Welfare Capitalism. Cambridge: Polity Press. Esping-Andersen, G. (1999) Social Foundations of Postindustrial Economies. Oxford: Oxford University Press. Jensen, Carsten (2007). Worlds of Social Services and Transfers: The Two Dimensions of Welfare State Regimes. Paper under review. Käärlälnen, Juha and Heikki Lehtonen (2006). The variety of Social Capital in Welfare State Regimes A Comparative Study of 21 Countries, European Societies, 8: 1, pp van Kersbergen, Kees (1995). Social Capitalism. A Study of Christian Democracy and the Welfare State. London and New York: Routledge. Knack, S. (2003) Groups, Growth and Trust: Cross-Country Evidence on the Olson and Putnam Hypotheses Public Choice 117 (304, December),

19 Kumlin, Stefan and Bo Rothstein (2005). Making and Breaking Social Capital: The Impact of Welfare State Institutions, Comparative Political Studies, 38: 4, pp Larsen, Christian Albrekt (2006). The Institutional Logic of Welfare Attitudes How Welfare Regimes Influence Public Support. Hampshire: Ashgate. Larsen, Christian Albrekt (2007). How Welfare Regimes Generate and Erode Social Capital. The Impact of Underclass Phenomena, Comparative Politics, forthcoming. Obinger, H., and Wagschal, U. (2001) Families of Nations and Public Policy, West European Politics, 24: 1, pp Oorschot, Win van and Wil Arts (2005). The Social Capital of European Welfare States: The Crowding Out Hypothesis Revisited, Journal of European Social Policy, 15: 1, pp Oorschot, Win van and Wil Arts (2006). Social Capital in Europe. Measurement and Social and Regional Distribution of a Multifaceted Phenomenon, Acta Sociologica, 49: 2, pp Nannestad, P., G.L.H. Svendsen and G.T. Svendsen (2007): 'Bridge Over Troubled Water? Migration and Social Capital'. Journal of Ethnic and Migration Studies (JEMS). In press

20 Nannestad, P. and G.T. Svendsen (2007): Institutions, Culture and Trust. Paper presented at the Quality of Governance conference (QoG), University of Gothenburg, Sweden (November 2005). Submitted. Paldam, M. (2004) The Nordic welfare state Success under Stress. European Journal of Political Economy, 20, Pierson, Paul (1994). Dismantling the Welfare State. Reagan, Thatcher, and the Politics of Retrenchment. Cambridge: Cambridge University Press. Pierson, Paul (1996). The New Politics of the Welfare State, World Politics, 48: 1, pp Pierson, P. (2000) Increasing Returns, Path Dependence, and the Study of Politics, American Political Science Review 94: Powell, M., and Barrientoes, A. (2004) Welfare Regimes and the Welfare Mix, European Journal of Political Research, 43 (1): Putnam, R. (1993) Making Democracy Work. Civic traditions in modern Italy. Princeton University Press, Princeton. Putnam, R.D. (2000) Bowling Alone. The Collapse and Revival of American Community. Simon & Schuster, New York, London, Toronto, Singapore

21 Rothstein, Bo (1998). Just Institutions Matter. The Moral and Political Logic of the Universal Welfare State. Cambridge: Cambridge University Press. Rothstein, Bo (2001). Social Capital in the Welfare State, Politics and Society, 29: 2, pp Rothstein, B. (2003) Social capital, economic growth and quality of government: The causal mechanism. New Political Economy, 8, Rothstein, Bo and Dietling Stolle (2003). Social Capital, Impartiality and the Welfare State: An Institutional Approach, in Marc Hooghe and Dietlind Stolle (eds.), Generating Social Capital. Civil Society and Institutions in Comparative Perspective. New York and Hampshire: Palgrave MacMillan. Shalev, Michael (forthcoming) Limits and Alternatives to Multiple Regression in Comparative Research, Comparative Social Research 25. Svendsen, G.L.H. and G.T. Svendsen (2003) On the Wealth of Nations: Bourdieuconomics and Social Capital. Theory and Society, 32 (5), Taylor-Gooby, Peter (1998). Markets and Motives. Trust and Egoism in Welfare States, Journal of Social Policy, 28: 1, pp Uslaner, Eric M. (2002). The Moral Foundations of Trust. Cambridge: Cambridge University Press

22 Uslaner, E.M. (2004) Where you stand depends upon where your grandparents sat: The inheritability of generalized trust. Working paper, University of Maryland, US. Vis, Barbara (2007): States of Welfare or States of Workfare? Welfare State Restructuring in 16 Capitalist Countries, , Policy and Politics, 35: 1, pp Weber, M. (1930 [1906]) The Protestant Ethic and the Spirit of Capitalism. Allen and Unwin, London. Wolfe, Alan (1989). Whose Keeper? Social Science and Moral Obligation. Berkeley, Los Angles and London: University of California Press

23 TABLES Table 1. GT in SD and CR Social democratic welfare states Denmark 64.1 Finland 56.8 Sweden 63.7 Average 61.5 Conservative welfare states Austria 31.3 Belgium 29.9 France 21.4 Germany 31.2 Italy 31.8 Average 29.1 Note: Generalised trust is measured as the percentage of respondents believing that most people can be trusted. Source: European Values Study

24 Table 2. Measures of welfare effort for SDR and CD SDR Public spending on cash benefits Public spending on old-age pensions Replacement rates Denmark Finland Sweden Average CD Austria Belgium France Germany Italy Average Source: OECD Social Expenditure Database and Economic Fact book (2006). Please refer to section 4 for details

25 Figure 1. A Model of the Creation and Persistence of Welfare States Cultural basis Mode of Trust Dominant Ideology Welfare Institutions

26 Table 3. Results of two-cluster iterative partitioning cluster analysis with k-means 1. cluster 2. cluster Countries Denmark, Finland, and Sweden Austria, Belgium, France, Germany, and Italy Trust variables Generalised trust (GT) Particularistic trust (PT) General welfare variables Total public cash benefits Public Old-age Pensions Replacement rates Familiaristic welfare variables Social care services Source: Generalised and particularistic trust (European Values Study); total public cash benefits, public old-age pensions, and social care services as percentage of GDP in the year 2001 (OECD Social Expenditure Database); replacement rates in case of unemployment in the year 2001 (OECD Tax-Benefit Models). End notes 1 Oorschot and Arts original index ran between one and three, but for presentational reasons we have multiplied the index by ten. 2 We use OECD s summary measure defined as the average of gross unemployment rates for two earnings levels, three family situations, and three durations of unemployment