WPS6447 Policy Research Working Paper 6447 Background Paper to the 2013 World Development Report Jobs, Wellbeing, and Social Cohesion Evidence from Value and Perception Surveys Frank-Borge Wietzke Catriona McLeod The World Bank Development Economics Office of the Senior Vice President and Chief Economist May 2013 Policy Research Working Paper 6447 Abstract Abstract: Recent events, including the Arab Revolutions once one passes the threshold from low to lower middle and protest movement of unemployed youths in OECD income countries, formal employment emerges as a countries, have contributed to the popular sentiment determinant of a range of outcomes relating to social that access to good jobs is an important driver of social cohesion, such as membership in social associations or cohesion. While economic dimensions of labor market levels of political activism. There are also indications of an outcomes are relatively well documented, evidence on the increasing association between work and life satisfaction link between social cohesion and employment conditions across higher and lower middle income countries. is still surprisingly scarce. This paper, a background report The paper concludes with a discussion of the study’s for the WDR 2013 on Jobs, presents descriptive evidence implications for emerging economies whose labor market that illustrates possible linkages between labor market and social institutions are still in transition. outcomes and social cohesion. The findings suggest that, This paper—prepared as a background paper to the World Bank’s World Development Report 2013: Jobs—is a product of the Development Economics Vice Presidency. The views expressed in this paper are those of the authors and do not reflect the views of the World Bank or its affiliated organizations. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at f.b.wietzke@lse.ac.uk. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Jobs, Wellbeing, and Social Cohesion Evidence from Value and Perception Surveys Frank-Borge Wietzke* and Catriona McLeod* JEL Codes I31, J01, 015 Keywords: labor markets, social cohesion, subjective wellbeing Sector Board: POV (Poverty Reduction) *London School of Economics, Department of International Development, Houghton Street, WC2A 2AE, UK. Email address of corresponding author (Wietzke): f.b.wietzke@lse.ac.uk Acknowledgements This paper was prepared as a background document for the 2013 World Development Report. Thoughtful comments and suggestions were provided by Ambar Narayan. Throughout we also benefitted greatly from discussions and practical support from the team. We are especially grateful for guidance and comments from Jesko Hentschel, Dena Ringold and Kathleen Beegle. However, we take full responsibility for any remaining omissions and mistakes. 1 I. Introduction Recent events, including the Arab Revolutions and protest movement of unemployed youths in OECD countries, have contributed to the popular sentiment that access to good jobs is an important driver of social cohesion. The OECD, in its recent report on social cohesion, argues that labor market outcomes are critical determinants of social stability, both because they influence the level and distribution of labor earnings and because jobs are critical loci of social interaction (OECD 2011: 154). In a similar vein, the World Bank’s 2013 World Development Report on Jobs identifies the link between labor markets and social cohesion as one of the central pillars of its multidimensional approach to jobs and development. While economic dimensions of labor market outcomes are relatively well documented, evidence on the link between social cohesion and employment conditions is still surprisingly scarce. Studies of insider-outsider conflicts over labor- market policies are typically limited to high income countries or they only consider economic outcomes (see among others Rueda 2005, 2006, Shayo 2009, Freeman 2008, Heckman and Pages 2000, Botero et al. 2004). On the other hand much of the cross country literature on social cohesion focuses on aggregate social outcomes – average levels of trust and civic attitudes or the quality of political institutions for instance (see Knack and Keefer 1997, 1995, Easterly et al. 2006, Easterly and Levine 1997, Rodrik 1998). However, few of these studies analyze specifically how social cleavages and distributional conflicts within and around national labor markets influence individual and aggregate outcomes. This paper, a background report for the forthcoming WDR 2013 on Jobs, presents descriptive evidence that illustrates possible linkages between labor market outcomes and social cohesion. The first link we consider involves associations between work and wellbeing. We focus on self-reported job preferences and the effect of unemployment on life satisfaction, assuming that these measures provide an indication of how labor marker-related priorities vary across countries at different stages of development. Second, we explore correlations between individual employment status and job quality and other more widely studied indicators of social cohesion, including self-reported levels of trust, political activism and social associations. Throughout, our analysis focuses on outcomes in middle and low income countries, given the relative dearth of information on social cohesion and labor market outcomes in this part of the world. However, for comparison purposes our study also report results from a set of high income countries. It is important to stress from the outset limitations of our data and analytical strategy. Data for this analysis are taken from the World Value Survey (WVS) project. To our knowledge this project provides the largest available number of 2 internationally comparable value and perception surveys, with a relatively good coverage of high, middle, and low income countries. However, the World Value Surveys are not dedicated labor force surveys and many of the employment categories used in this analysis are not as carefully defined and tested as in other data bases on labor market outcomes. More importantly, the surveys used in this paper do not enable us to present causal support for the association between labor market outcomes and individual experiences of social cohesion (in particular we cannot distinguish whether individuals who are out of work or in lower quality job report lower levels of social and subjective wellbeing because of their employment situation, or because they had poorer subjective and social outcomes to start with). Nonetheless, we hope that non-causal statistical associations presented in this paper can offer initial indications or trigger future research on how labor market outcomes influence social cohesion in different parts of the world. II. Jobs and social cohesion The literature has employed multiple, often loosely defined concepts of social cohesion, incorporating, to various extents, notions of social inclusion and exclusion, participation, and social and political rights. Yet, one theme that runs through most contributions involves concerns about the undermining of the social fabric by economic and social transformations. In particular in the last decades these concerns have often been related to rapid economic globalization and its potential consequences for job security, wage inequalities, and social and cultural distances between groups with uneven ties to the labor market (Norton and De Haan 2012, OECD 2011). This paper follows the WDR’s framework on social cohesion in that it emphasizes the way societies and groups manage possible collective action problems arising from economic and social transformations in and around national labor markets. 1 In addition, we emphasize the distribution of social wellbeing outcomes within societies. 2 We believe that this definition provides a useful starting point for highlighting different manifestations and experiences of social cohesion among various groups and individuals with different levels of attachment to the labor market. With this definition of social cohesion in mind it is possible to think of several ways in which labor market outcomes may contribute to social stability. Starting from 1 This follows Woolcock (2011) who defines social cohesion as “the capacity of societies (not just groups, networks) to peacefully manage collective action problems.” 2 This brings our definition of social cohesion closer to that of the OECD (2011), which describes a society as cohesive if it “works towards the well-being of all its members”. 3 the individual level, it can be hypothesized that jobs can contribute to social cohesion through their effects on personal wellbeing, identities, and political and social preferences. 3 Individuals who perceive that they are paid fairly and not disadvantaged in their search for jobs are less likely to disengage from the political and social institutions of their communities. Likewise, workers who feel that their job prospects depend at least partially on the overall performance of a nation’s economy may sense a vested interest in the political and economic institutions of their country. This interdependence of interest can align personal with societal preferences and contribute to a stronger sense of national identity and fewer conflicts within the labor market. It has also been suggested that jobs may contribute to social cohesion through their effects on inter-personal interactions and networks. Sociologists of development have long argued that people’s economic activities are underpinned by people’s social and professional ties (“social capital”). While the direction of causation from social capital to professional success is not usually easy to determine, people in higher paying or upwardly mobile employment situations often have a higher probability of entertaining more extensive and diverse networks across social contexts and communities (“bridging” social capital). In contrast, individuals in lower paying jobs or subsistence farming are more likely to have interactions that are restricted to their immediate community or kinship group (“binding” social capital; Woolcock and Narayan 2000, Baker 2001, Katzman 2005, see also Granovetter 1973, Coleman 1988). The flip side is that jobs are often closely associated with processes of social transformation and stratification that can, ultimately, lead to social tensions. Sociologists have long viewed work-related identities as the main determinant of social differentiation in modern societies. In particular the Weberian tradition, best exemplified by the work of sociologists Erikson and Goldthorpe, views the type of job a person holds not only as the primary source of his or her social identity but also of his or her social status (Goldthorpe 1992, Breen and Rottmann 1995). In settings with high levels of social segregation these identities may be reinforced by the breakdown of interactions across status and group divides (Bertaux and Thompson [ed.] 1997). The consequence may be a permanent reduction in social and employment-related opportunities among disadvantaged groups, as well as a growing sense of alienation across group divides that can ultimately undermine social cohesion (see Box 1). Political scientists and economists, albeit less concerned with notions of economic class, have also pointed to risks for social cohesion around worker- employer relations. In most countries workers and employees represent the largest and most vocal groups in the electorate. This creates natural incentives for policy makers to prioritize the interests of labor market insiders over those of labor market outsiders (see for instance Rueda 2005, 2006, Shayo 2009). The result may be employment 3 For examples see Dudwick 2012. 4 regulations that greatly enhance job quality, security and incomes for individuals who are already in steady work. Yet, this may be achieved at the price of considerably higher barriers to entry into the labor market for individuals with lower skill sets and weaker social and political associations (Botero et al. 2004, Heckman and Pages 2000). 4 Box 1. Social exclusion and the underclass Sociologists working on social inequalities in high income countries have long been concerned about the emergence of a so-called “underclass”. The rise of the underclass is typically explained by structural shifts in the economy that lead to a reduction in employment opportunities and real wages for low and semi-skilled workers. However, underclass status is often described as a more general sociological condition with causal origins well outside the labor market. Widely cited causes for the reproduction of underclass status include processes of urban segregation and ‘ghettoization’, social and cultural discrimination, and the breakdown of conventional family structures. With respect to social cohesion, underclass status is often associated with formal and informal processes of disenfranchisement: members of the underclass are excluded from social and economic institutions through informal forms of discrimination and they are less likely to engage in the type of productive and constructive collective action that would better their situation and integrate them more permanently in society. This may result in mutual antagonism between the excluded and included and it may explain why the interests of the poorest are not always well represented in the politics of their countries (see Wilson 1997, Anderson 1999, see also Grusky and Kanbur 2006, Durlauf 2002, Morgan et al. [eds.] 2006. While stratification around labor relations may cause social or political conflict, it appears that this factor alone does not always undermine social cohesion. Concerns about the break-down of social cohesion around labor relations have been motivated primarily by recent macro-economic transformations, such as a reduction in the share of labor income in total value and increased competition due to technological change and globalization (OECD 2011: 70ff).This is in contrast to earlier years of the post war period when labor markets and societies were seen as functionally more integrated. For example, looking back to the period before the oil shock of the 1970s and the ensuing economic adjustment processes of the 1980s, most observers paint an almost romantic picture of labor markets and class relations, when economic growth and steady increases in real wages yielded social and economic benefits for wide sections of society (see for instance Rodrik 1997, Katzman 2005). 4 Note however, that the cross country literature has not produced consistent evidence on the economic consequences of labor market regulations. Studies that associate worker protection to economic efficiency losses include Heckman and Pages 2000 and Botero et al. 2004. Other cross country studies find no or inconsistent effects of worker protection on economic performance (Freeman 2008, 2009, Forteza and Rama 2006…). 5 Recent events in North-African, Middle Eastern and some European and Latin American countries illustrate that conflicts over labor relations primarily arise in times of economic or political transformation: many of these societies maintained seemingly high levels of social stability for long periods of time, in spite of relatively large underlying differences in economic opportunities. However, these latent conflicts became politically ‘salient’ in the light of changing macro-economic and political outlooks. Additionally, the extent and duration of conflicts in times of economic transformation can depend on the wider institutional and social context of a society. Economic theory suggests that societies with more consensual institutions will be better able to absorb conflicts between winners and losers of economic and political transformations (Fernandez and Rodrik 1991, Alesina and Drazan 1991). While empirical evidence on the specific link between national institutional contexts and labor-related conflict is still relatively scarce (see below), there is at least indirect empirical support for this hypothesis. Rodrik (1999) for instance, finds that societies with more consensual institutions generally experience fewer and shorter periods of economic crisis (see also Easterly et al. 2006). Other studies document more general associations between overall levels of social cohesion and slower moving institutional attributes of a country, including initial levels of economic inequality, ethnic fractionalization or the type of welfare state regime adopted by a society (Knack and Keefer 1997, Easterly 2001, Easterly and Levine 1997, Easterly et al 2006, Larsen 2007, Rothstein and Uslaner 2005). 5 III. Existing evidence and gaps in the literature Even though the link between labor market policies and economic outcomes is relatively well documented in industrialized societies (for overviews see Lindbeck and Snower 2002, Freeman 2008), evidence for many parts of the developing world is less clear. Cross country comparisons that show that labor market regulations lead to increased inequalities in the workforce include Heckman and Pages (2000) and Botero et al. (2004). However, other studies find no or inconsistent linkages between worker protection and economic performance. For instance Forteza and Rama (2006) find no effect of Employment Protection Legislation (EPL) on the depth and durability of economic crisis in a sample of over 100 transition economies. But they show that the extent of labor organization deepens economic crisis and slows down the adjustment process. Freeman (2009) finds no consistent evidence between the level of labor regulation and growth rates (see also OECD 2011 and Calderon and Chong 2005, cited in Freeman 2009). 5 These studies variably measure social cohesion by national levels of self-reported trust, civic attitudes, social associations, political stability, civic liberties and violent conflict. 6 Once one moves beyond purely economic effects of employment polices to political and social outcomes, evidence becomes even scarcer. Studies of social transformations associated with recent macro-economic changes in lower and transition economies are mostly limited to isolated case or country studies that are hard to compare across societies. 6 Likewise, most studies that evaluate conflicts over employment policy between labor market ‘insiders” and “outsiders” are limited to high income countries (see for instance Rueda 2005, 2006, Shayo 2009). 7 To be sure, the dearth of evidence about the political and social outcomes from employment policies in lower income and transition economies also reflects data limitations. In particular value and perception surveys that provide the bulk of information for studies of political conflicts over labor market regulations in high income countries generally do not report employment-related information at the same level of detail in developing countries. Yet we believe that even at current levels of data availability, important insights can be gained from more careful analysis of the link between work and social wellbeing. The first area where we see scope for new contributions involves inequalities in social outcomes between different groups in the labor market. While valuable progress has been made in the harmonization and documentation of internationally comparable indicators of social development, most cross country studies and data collection exercises still base their comparisons on national averages of social and institutional development indicators. 8 It is clear that this format is less useful for identifying differences in outcomes across individuals and groups with different attachment to the labor market; and it limits our ability to assess claims about the link between labor market outcomes, social exclusion and insider politics often made in the literature. For example, while it could be hypothesized that some countries attain relatively high levels of average social wellbeing because of policies that protect the interests of a majority of labor market “insiders” against the concerns of smaller groups of labor market “outsiders”, average measures of social development would conceal resulting inequalities in social outcomes. As a result these measures are less helpful for evaluating competing arguments about social and political tensions over labor market policies. 6 See for instance Wietzke 2009 for a review of evidence from Latin America. 7 For an exception see Blanchflower and Freeman 1997 who report preferences on economic and employment policies in a sample of Eastern European transition countries. For cross country studies on preferences over inflation see Shiller 1996, Di Tella et al 2001. Boeri et al. (2001). Di Tella and Mc Culloch 1996 and Luttmer 2001 report evidence on attitudes toward welfare state policies and redistribution. 8 Examples include the Indices of Social Development (ISD) hosted by Institute of Social Studies in The Hague (http://www.indsocdev.org/) and the World Governance indicators data base by Kaufman et al (2010). See also the literature on social cohesion cited above. Many of these studies rely on the same value and perception surveys we use below. 7 A more fundamental question is whether concerns about job quality and social cohesion can be easily generalized across countries at different stages of economic development. It is well documented in the cross country literature on life satisfaction that incomes and other material inputs to a person’s quality of life diminish in importance as societies grow richer. 9 Increases in incomes only raise wellbeing up to a level where households are, on average, able to comfortably meet their basic material needs. However, as societies grow wealthier, added incomes, including from wage earnings, only have a small (or no) effect on subjective wellbeing; while other non-material aspects of people’s lives become more important (such as social status, the quality of public institutions and personal relations, the state of the environment, etc. See also Inglehart and Welzel 2005). These variations in the determinants of wellbeing could interact with more general attributes of national labor markets, in ways that shape both work-related experiences of social interactions and job preferences. For instance, weaker formal safety nets and social protection policies in low income countries may be accompanied by stronger reliance on family or community-based support networks, with obvious consequences for the social experience of unemployment. Likewise, lower skilled, but safe and adequately paid factory jobs may provide a significant boosts to life satisfaction in settings with high levels of poverty and economic informality. 10 Yet such jobs would surely do little to raise wellbeing in higher income countries with a more diversified and better skilled work force. Public priorities and social interactions in and around labor markets thus vary along with economic and institutional contexts, with clear implications for analysis and policy. 9 This follows work by the Economist William Easterlin and is commonly known as the Easterlin Paradox (Easterlin 1974, Easterlin et al 2010). 10 More generally, unemployment may be a less meaningful category for studying differences in social and economic participation in settings with high levels of economic informality, given that most people are engaged in some sort of income-generating activities. For instance, unemployment, defined by temporal inactivity during the transition from one job to another may be more common among the relatively well-off, who can rely on formal insurance or other sources of household income (savings, capital returns etc.) 8 IV. Objectives and empirical strategy The aim of this paper is to offer a first explorative analysis of the extent of work-related inequalities in social wellbeing and social cohesion across high, middle, and low income countries. Our analysis focuses on two things. i. First we document self-reported job preferences and differences in self-reported (subjective) wellbeing between the employed and the unemployed. Following our discussion above we compare self-reported job preferences as an indicator of possible differences in priorities for labor market policies in high and lower income countries (see below). In addition, we hypothesize that the extent of variations in life satisfaction across different groups in the labor market may be seen an indirect measure of the inclusiveness - or lack thereof - of a country’s social and political institutions. To this end we also discuss how levels of subjective wellbeing for the employed and unemployed co-vary along with other economic and institutional attributes of a society. ii. As a second step we explore whether or not uneven levels of wellbeing between the employed and unemployed translate into differences in social activism and other outcomes that could serve as more direct proxies for the level of social cohesion in a society. Our main indicators are self-reported levels of trust in people, a variable that is often used in the literature to approximate levels of social cohesion in a society (see for instance Knack and Keefer 1997, Larsen 2007). Other outcomes include people’s membership in civic associations, levels of political activism and attitudes towards formal governance institutions. We study these outcomes to gain an impression how effectively groups with different links to the labor market are able to participate in the social and political institutions of their countries. Again, our analysis distinguishes between higher and lower income countries. We use data from the fifth round of the World Value Survey (2005-6), to our knowledge the largest freely available and internationally comparable data base with information on social wellbeing outcomes. 11 Surveys include a large number of questions on social wellbeing and associations, political attitudes and other outcomes that can be used as indicators (direct or indirect) of social cohesion. The surveys are coordinated by an international network of researchers from 54 nations, which ensures comparability of questionnaires and interview protocols across countries. Since we are interested in the gradation in work and social cohesion outcomes across national income levels we divide the sample according to World Bank country and lending groups (see Table 1). While the total sample size includes over 82,000 individuals, we 11 See http://www.worldvaluessurvey.org/ for data and documentation. 9 focus only on members in the labor force (to ensure more meaningful estimates of the welfare effect of joblessness and other job-related categories). Remaining sample sizes for each country group are reported along with our regression outputs below. Table 1. Countries by sub-sample (1) (3) (5) (7) Country income High income Upper middle Lower Low income group (based on income middle World Bank / WDI) income Countries in sample Australia Andorra Egypt Burkina Faso Great Britain Brazil Georgia Ethiopia Canada Bulgaria Ghana Mali Finland Chile Guatemala Rwanda France China India Germany Colombia Indonesia Italy Cyprus Iraq Japan Hong Kong Moldova SAR, China Netherlands Iran Morocco New Zealand Malaysia Ukraine Norway Mexico Vietnam Poland Peru Zambia South Korea Romania Slovenia Russia Spain South Africa Sweden Serbia Switzerland Thailand Trinidad Turkey Uruguay As much as possible, our analysis seeks to capture the ‘pure’ effect of unemployment or job quality. As a consequence most of our estimates rely on individual level regression that control for other personal attributes that may influence wellbeing and personal experiences of unemployment. The underlying estimation model is of the following form: Yic=α1 + β2 employment status or job quality ic+γXi +ξc + εic (1) where the dependent variable Yic stands for a range of outcomes, such as life satisfaction, self-reported trust or social and political activism (see below). Suffixes i and c denote individuals and countries, respectively. 10 The main explanatory variables in the model are employment status and job characteristics. Employment status is measured by an indicator that identifies unemployed individuals (1=unemployed). Job characteristics are measured by a summary index, based on information whether respondents perceive their work to involve more manual or cognitive and more routine and creative tasks, and how much independence they have at work. Each variable is measured on a 10-point scale with higher values for more self-determined tasks. The index is the simple average of these scores. While these measures of job characteristics were available for a large sample of countries within the WVS, it is important to highlight that they may be a less accurate indicator of job quality in lower income countries. 12 Where appropriate our estimates also account for part-time and self-employed, lower and unskilled workers and farmers. Unless noted otherwise, all estimates include a vector of individual and household level controls Xi. Similar to other specifications encountered in the literature these controls include age, age squared, family status (married, widowed, or divorced), level of education (secondary or higher), number of children, gender, and an indicator for persons who actively practice a religion. Our estimates also control for self-reported household income in an attempt to filter out the non-monetary dimensions of joblessness and job quality. ξc is a vector of country fixed effects that account for unobserved country-level differences and determinants of social wellbeing. It is important to highlight a number of shortcomings of our data and estimation strategy. Our data are based on a simple cross section of individuals and do not include information on personal job and life histories. This means we are unable to distinguish whether differences in life satisfaction and social wellbeing reported by people in different employment contexts result from experiences in the labor market, or whether they reflect initial differences in individual circumstances and attitudes. For example, unemployed individuals may be less likely to trust others or participate in social and political activities because of the social and economic costs of joblessness. But they may also be more likely to be unemployed because they lacked wide spread social networks in the first place. Our individual level estimates control for self-expressed work ethics in an attempt to filter out some initial differences in work attitudes (more specifically we use indicators that identify respondents who feel that work is “very” or “somewhat” important). However, this procedure cannot account for all the differences in personal preferences and life histories that may influence a person’s current work-related and social outcomes. We thus refrain from assigning causality to any of our results. 12 For example people in lower qualified factory jobs in low income countries may place more importance on job security, cleanliness and safety of the work place or transportation. However, unfortunately information on these job attributes was not available in the WVS 11 In addition, our results may be influenced by more general interactions between labor market outcomes and levels of cohesion in a society. For example, employers in societies with greater generalized levels of trust may be less likely to discriminate on the basis of social stereotypes. This implies that the characteristics of those who are unemployed, as well the social experiences associated with joblessness, may vary across countries with different initial levels of social cohesion. 13 In spite of these concerns we hope that also the simpler comparisons of work- related differences in social and subjective wellbeing presented in the following sections can provide initial indications of the interaction between social cohesion and labor market outcomes in different parts of the world. V. Jobs and subjective wellbeing Recent events in the Arab World and other transition economies suggest that differences in wellbeing between those in and out of work pose increasing challenges to social stability in emerging economies. Yet we know surprisingly little about how jobs affect wellbeing, once we leave the relatively small group of well-researched high income countries. In high income countries the link between unemployment and low life satisfaction is fairly well established. A ground-breaking study of unemployment in Britain reports that the experience of joblessness “depressed well-being more than any other single characteristic, including important negative ones such as divorce and separation” (Clark and Oswald 1994). Subsequent research indicates that these effects are typically robust to the inclusion of controls for income, suggesting that a large part of the wellbeing effect of unemployment works through non-monetary channels (for overviews and evidence see Di Tella et al 2001, 2003 and Frey et al. 2008). There are also indications that the causal link flows from unemployment to subjective wellbeing. For instance, research carried out on German panel data suggests that the same individuals experience considerable reductions in life satisfaction when they lose their job (Winkelmann and Winkelmann 1998, Frey 2008: 47). Outside of the group of high income countries evidence tends to be limited to a relatively small group of middle income and transition economies. For example, estimates on a pooled sample of Latin American economies finds a clear and unambiguous effect of unemployment on reported life satisfaction, even when a range of personal and household attributes are accounted for (Graham 2008, see also Graham and Pettinato 2002). Similar results have been presented for Russia and formerly socialist economies in Eastern Europe (Blanchflower and Freeman 1997, 13 However, the country fixed effects in our estimation model will account for some of these variations. 12 Eggers et al. 2006). Yet, to our knowledge, evidence is much scarcer in countries in lower income countries with less developed labor markets. Based on our earlier discussion of linkages between national incomes and average wellbeing we should expect that job preferences, and the life satisfaction derived from work would vary considerably across countries with different income levels. This is also borne out by data from the World Value Surveys. When asked what they value most in a job, people in high income countries are much more likely to report that they value work that they perceive as “important” (Figure 1). This is in contrast to low income countries, where a much higher share of respondents reports a preference for jobs that generate higher earnings. There are also indications that job preferences interact with other outcomes in national labor markets. For example, people in countries with higher unemployment rates are more likely to report a preference for safe jobs and less likely to seek work they perceive as ‘important’. This suggests that, in addition to depressing overall wellbeing, higher unemployment rates also have a negative effect on people’s aspirations to carry out meaningful work. However, the correlations between national unemployment rates and job preferences in our sample are driven by a relatively small group of countries with very high unemployment rates. Figure 1: Preferences for jobs GNIpc and preference for 'high earnings' GNIpc and preference for 'important work' .8 .5 .4 .6 .3 .4 .2 .2 .1 0 0 10000 20000 30000 40000 0 10000 20000 30000 40000 GNI per capita (constant 2000 US$) GNI per capita (constant 2000 US$) (mean) imp_income Fitted values (mean) imp_job Fitted values Source: WVS 2005-7 and WDI 2005 data Source: WVS 2005-7 and WDI 2005 data 13 unemployment and preference for safe jobs unemployment and preference for 'important work' .5 .5 .4 .4 .3 .3 .2 .2 .1 .1 0 5 10 15 20 0 5 10 15 20 Unemployment, male (% of male labor force) Unemployment, male (% of male labor force) (mean) safe_job Fitted values (mean) imp_job Fitted values Source: WVS 2005-7 and WDI 2005 data Source: WVS 2005-7 and WDI 2005 data Source WVS wave 5, full sample, selected responses to question v48 The WVS is not a dedicated labor force survey and there are some concerns about the quality of the definition of underlying employment categories. 14 For example unemployment may be measured with higher measurement error in lower income countries or unemployment may be a less meaningful category in settings where poorer households cannot afford to forgo income generating activities. However, in spite of these caveats we observe interesting cross-country patterns in the association between life satisfaction and employment status when we turn to individual-level regression estimates (Table 2). In the case of unemployment we see a relatively clear gradation across countries at different income levels. The coefficient of unemployment is largest in high income countries (-0.64) and it is reduced by a comparatively small amount (to -0.51) as one moves to high and middle income nations (Panel A). The coefficient then drops by almost two-thirds in lower middle income countries (-0.18). In low income countries, here exclusively represented by economies in Sub-Saharan Africa, the effect of unemployment becomes entirely insignificant and the sign of the coefficient actually reverses. 14 To address these issues we also estimate the wellbeing effect of unemployment using data from the 2006 European Bank Reconstruction and Development’s Life in Transition Survey (LITs). Among several value and perception surveys reviewed for this study, the LITs has the most detailed information on unemployment and job histories. We used this information to carry out robustness tests with alternative definitions of unemployment and more detailed controls for other job categories. These tests did not generate qualitatively different results, relative to the findings reported below. Results are available on request from the authors. 14 Table 2. Links between employment status, job quality, and life satisfaction (1) (3) (5) (7) Panel A Life satisfaction and unemployment High income Upper middle Lower Low income income middle income Unemployed -0.637*** -0.512*** -0.177** 0.163 (0.075) (0.062) (0.077) (0.123) Self employed 0.019 -0.024 -0.041 0.060 (0.054) (0.055) (0.064) (0.099) Lower skilled -0.267*** -0.138** -0.121* -0.064 (0.064) (0.055) (0.063) (0.130) Part time -0.065 -0.173** -0.099 0.178 (0.053) (0.068) (0.077) (0.171) Farmer -0.303* -0.060 0.061 -0.044 (0.162) (0.090) (0.095) (0.117) Income 0.101*** 0.215*** 0.358*** 0.392*** (0.009) (0.011) (0.015) (0.021) Saved part of income 0.407*** 0.393*** 0.446*** 0.294*** last year (0.039) (0.044) (0.063) (0.084) Country fixed effects Yes Yes Yes Yes Sigma 1.716*** 2.315*** 2.398*** 2.070*** (0.017) (0.016) (0.020) (0.029) Constant 7.759*** 7.635*** 7.854*** 3.329*** (0.210) (0.229) (0.276) (0.480) Observations 11241 17017 11253 3566 15 Panel B Life satisfaction and Job quality High income Upper middle Lower Low income income middle income Job quality 0.066*** 0.047*** 0.078*** 0.003 (0.010) (0.012) (0.026) (0.016) Self employed 0.007 -0.083 -0.048 0.046 (0.057) (0.059) (0.105) (0.074) Lower skilled 0.025 -0.169** -0.189 -0.093 (0.053) (0.071) (0.147) (0.086) Part time -0.156 -0.202*** -0.287** -0.110 (0.097) (0.070) (0.126) (0.080) Farmer 0.087 0.162 -0.067 -0.255** (0.157) (0.105) (0.160) (0.103) Income 0.081*** 0.186*** 0.347*** 0.387*** (0.009) (0.012) (0.024) (0.018) Saved part of income 0.345*** 0.394*** 0.278*** 0.464*** last year (0.040) (0.047) (0.083) (0.071) Country fixed effects Yes Yes Yes Yes Constant 7.059*** 5.567*** 4.296*** 3.603*** (0.242) (0.267) (0.566) (0.370) Observations 8719 13183 2765 7085 Estimates based on World Value Survey Wave 5. The dependent variable is self-reported life satisfaction measured on a 10pt scale, with higher values denoting higher life satisfaction. Coefficients are calculated with a tobit model to account for possible over reporting on the highest score of the scale. Results were not qualitatively different when we used OLS or ordered Probit models. All estimates control for age, age squared, family status (married, widowed, or divorced), level of education (secondary or higher), number of children, gender, religious attitudes, and work ethics (work is “very” or “somewhat” important). *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors are in parentheses. Differences in the effect of job characteristics are less clearly differentiated across country income groups. However, here also we find that job attributes, as measured by our data, have no independent effect on subjective wellbeing in the poorest economies (Panel B). Also the control for lower skilled workers in the model with unemployment becomes indistinguishable from zero in the poorest countries in our sample (Panel A). Yet, this result may also be explained by the fact that our measure of job characteristics may give less accurate descriptions of the variability of people’s working conditions in lower income countries. In addition there are other indications that work-related priorities vary across countries at different stages of economic development. Considering the control for household income we see that income contributes increasingly more to life satisfaction with decreasing levels of national income (the effect of income already 16 becomes quite sizeable in lower middle income countries). This finding may reflect the fact that in setting with less differentiated labor markets income may still be a more important determinant of people’s subjective life evaluations than the employment categories in the WVS. Subjective wellbeing and other institutional contexts Even though the documented cross-country associations between average incomes, work, and wellbeing are not entirely surprising, they do raise a potentially puzzling question about social cohesion in higher income countries: Why do some wealthier nations with larger employment-related differences in subjective wellbeing consistently rank among the most stable and cohesive societies in the world? 15 The answer to this question appears to involve interactions between average incomes, social cohesion and the wider institutional and policy contexts of a society. Countries with higher levels of average income often maintain more elaborate safety net and social protection mechanisms. These can mitigate tensions between winners and losers of economic transformations and raise average social and subjective wellbeing. In addition, one would expect that more general institutional and social attributes of a society, such as the overall quality of public institutions and social relations, would have effects on wellbeing that are shared more widely among individuals in different positions in the labor market. These points also emerge when we examine levels of wellbeing for the employed and unemployed in relation to the general economic and institutional context of various countries. The top left graph in Figure 2 maps levels of wellbeing of the employed and unemployed across quintiles of Gross National Income (the graphs in this panel do not control for other country level and individual attributes). As seen before, differences between the two groups are smallest in the poorest countries, but the gaps increase rapidly even at comparatively low levels of average income. Yet, these changes in relative levels of wellbeing are accompanied by overall higher levels of absolute welfare for both groups in wealthier societies (even though life satisfaction slightly decreases for the unemployed in the highest income quintiles). At least to some extent, the benefits of rising material living standards appear to be shared between the employed and unemployed. 15 See evidence above. In particular Larsen 2007 and Rothstein and Uslaner 2005. 17 Figure 2. absolute levels of life satisfaction and country-level attributes Subjective wellbeing of the employed / unemployed Subjective wellbeing of the employed / unemployed ranked by quintiles of per cap GNI ranked by quintiles of rule of law index 8 8 6 6 4 4 2 2 0 0 1 2 3 4 5 1 2 3 4 5 mean of lsatisEMP mean of lsatisUNE mean of lsatisEMP mean of lsatisUNE Subjective wellbeing of the employed / unemployed Subjective wellbeing of the employed / unemployed ranked by quintiles of government effectiveness index ranked by quintiles of labour regulation index 8 8 6 6 4 4 2 2 0 0 1 2 3 4 5 1 2 3 4 5 mean of lsatisEMP mean of lsatisUNE mean of lsatisEMP mean of lsatisUNE Subjective wellbeing of the employed / unemployed Subjective wellbeing of the employed / unemployed ranked by quintiles of gov' transfers ranked by quintiles of national unempl rate 8 8 6 6 4 4 2 2 0 1 2 3 4 5 0 1 2 3 4 5 mean of lsatisEMP mean of lsatisUNE mean of lsatisEMP mean of lsatisUNE Red bars denote average life satisfaction on a 10 pt scale for the unemployed and blue bars for the employed. GNI and unemployment rates are from the World Bank WDI data base. Rule of law and government index from Kaufman et al (2010). Government transfers from Gwartney, J. et al. (2010). labor market regulation index denotes the labor laws index from Botero et al. (2004). Similar trends emerge when we consider more general indicators of the quality of a nation’s institutions. Mapped against an index of government accountability (rule of law) and government effectiveness the level of overall life satisfaction increases for both groups, even though relative differences in welfare are more marked in countries 18 with higher governance quality (see Kaufman et al 2010 for data sources and descriptions of the indices). Since it is possible that levels of institutional development are simply proxies for national income levels, we performed additional estimations of levels of life satisfaction for the unemployed and employed that controlled for GNI with the same set of individual controls used in our previous regressions as well as an index denoting the share of government transfers in the Budget (from Gwartney, J., J. Hall, et al. 2010). These regressions also indicated a positive association between institutional quality and the level of life satisfaction for both groups (see Table 2, Annex). Somewhat surprisingly, differences in wellbeing between the employed and unemployed do not widen significantly when we consider other outcomes that are more directly related to labor market outcomes and policies. Mapped against an index of the depth of labor market regulations produced by Botero et al.(2004), the share of government transfers in the national Budget and the national unemployment rate, we observe more or less converging trends for both groups. This finding may reflect that weaker social protection policies and high unemployed often lead to a heightened sense of insecurity among the unemployed as well as those in work. This finding is consistent with earlier results that suggest that reforms of employment-related safety nets are often experienced in similar ways by the employed and the unemployed. For example Di Tella et al. (2003) show that increases in unemployment benefits in Europe from 1975 to 1992 did not narrow differences in reported happiness between the unemployed and the employed because absolute levels of wellbeing grew for both groups (Di Tella et al. 2003). The authors hypothesize that this result has to do with the reduction of the financial risks of unemployment for both groups. While these results clearly merit more careful investigation they do point to one potentially important lesson about the link between social cohesion and jobs: social transformations and distributional conflicts in national labor markets do not usually occur in isolation. They are deeply embedded in, and influenced by, the wider social and institutional context. This also has implications for economic and social policy, which we will discuss after the following section. VI. Jobs and social participation The previous section has illustrated how economic contexts influence the level of subjective wellbeing that people around the world derive from work. However, it has provided no indication of how effectively different groups in the labor market participate in the social and political institutions of their societies. Are people who are unemployed or in more fulfilling forms of work also more or less likely to participate 19 in the social and political institutions of their countries? To address this question we turn to social outcomes that can be considered more direct indicators of the quality and inclusiveness of social relations in a society. Key variables analyzed in this section include the number of social associations entertained by an individual, levels of political activism and participation, and the general levels of trust experienced by people. It is possible to envisage a number of reasons why different groups in the labor market have different capacities to participate in the social and political institutions of their societies. From an aggregate perspective differences in the ability to participate in collective decision making over labor market policies often arise from organizational problems. In particular, collective action theory, often used by economists and political scientists to explain employment policies that raise high barriers of entry into the labor market, predicts that the unemployed will find it harder to have their voice heard in society, because they represent a more diffuse group than individuals who are already in work (and possibly organized in unions). 16 In settings where wage levels and labor market policies reflect institutional negotiations between workers and employers this may lead to outcomes where the benefits of labor market reforms are very unevenly distributed between labor market ‘insiders’ and ‘outsiders’. The former group will tend to negotiate employment regulations that protect those already in work (such as raising regulatory obstacles to the firing of workers). But this is achieved at the cost of higher barriers to employment for those who are out of work (see above). 17 Organizational challenges among the unemployed can be exacerbated by social institutions that relate more directly to our indicators of social cohesion. As noted before, high levels of social and residential segregation can weaken social associations across status divides and contribute to a sense of isolation and disenfranchisement among disadvantaged groups. This can be accompanied by negative perceptions of the effectiveness of democracy and public institutions (Altindag and Mocan 2010), which may further reduce incentives for the unemployed to engage in political action. In addition, because identities and behaviors are often reproduced within communities and households, social and family background can become a very important determinant both for an individual’s access to the labor market and to wider social and political institutions (Wilson 1997, Morgan et al. [eds.] 2006, Durlauf 2002). We could also expect that the psychological costs associated with job losses influence the social and political activities of the unemployed. Another WDR 16 Classic references include Olson 1971 and Olson 1982. 17 Insider-outsider politics can be particularly pervasive in settings where advantaged groups form political coalitions with outsiders on cultural issues, such as religion, family and gender policies, or notions of national identity (see for instance Shayo 2009). 20 background paper that tracked levels of community participation in Indonesia found that men and women who became unemployed between 2000 and 2007 were less likely to be involved in community activities (controlling for personal attributes). In some locations this result was observed in spite of a general increase in community activities, indicating that individual and household effects of unemployment can influence personal social outcomes, independent of social and community contexts (Giles et al. 2012). More qualitative evidence also suggests that job loss can lead to a breakdown of social associations. For example, a man who recently became unemployed in Serbia explained, “I automatically lost everything. I lost any freedom and power I had. Everything was lost.” (Petesch 2012). The data from the WVS do not enable us to disentangle the specific factors that influence individual attitudes and levels of participation. In particular, the surveys do not record information on social contexts or parental background and we do not have information on people’s social attitude and activities in earlier periods of employment or education. However, in spite of these limitations, and keeping in mind the limitations of our measures of job status and quality in lower income countries, there are indications that levels of social and political activism vary quite considerably across individuals who differ with respect to their employment status or job quality. Figure 3. Links between employment status, job quality and perceived social status Unemployment and lower class status Job quality and upper class status probit coefficients probit coefficients .7 .2 .6 .5 .1 .4 .3 .2 0 .1 0 -.1 -.1 OECD high middle low middle low inc. OECD high middle low middle low inc. incom_grp incom_grp 1 min95/1 max95 1 estimate 1 min95/1 max95 1 estimate The vertical axis shows the probit coefficient and 95% confidence interval of individuals’ self-reported class status. In the case of the unemployed the outcome of interest is self-reported membership in the “lower class”. In the case of job quality the outcome is membership in the “upper or upper middle class”. The analysis controls for the same set of controls listed at the bottom of Table 2. Note that the graphs use different scales. If we take people’s self-assessment of their “social class” as a summary measure for status-related social experiences, we see a comparatively clear association between social wellbeing and people’s employment status (Figure 3). In all sub-samples the unemployed are much more likely to place themselves in the lowest class. This finding is in fact quite robust across all the country categories and 21 also holds in the lowest income countries. However, the effect is smaller and less robust in the case of job characteristics (more specifically we consider the association between self-reported job attributes and perceived upper or upper middle class status). In particular, job quality does not appear to predict class status in lower middle income countries. Similar results emerge when we consider people’s associational life. Figure (4a) presents correlations between people’s employment status and their participation in recreational, religious, or political associations as indicators of the density of an individual’s social networks, showing much lower outcomes among the unemployed - i.e. those out of work generally participate in fewer associations. This effect is again much stronger in high income countries and it is only borderline significant in the poorest nations in our sample. There are also indications that people in less routine and more autonomous jobs participate more frequently in social associations. However, this result does not hold for lower income countries (Figure 4b). Turning to people’s political activities we see some signs that the unemployed are often less engaged in the political institutions of their country. Overall the unemployed are less likely to participate in demonstrations or to sign petitions. Yet, these outcomes are less clearly differentiated between higher and middle income countries (the effect is again statistically insignificant in the case of low income countries). People in higher quality jobs tend to be more politically active in all sub- samples but the lowest income countries. 18 There are however, indications that not all dimensions of social cohesion are perceived differently by individuals in different employment situations. In particular, we find that differences in self-reported trust in other people – one of the key indicators for the level of social cohesion in the cross country literature – only exist across employment categories in high income countries. In low income and transition economies levels of trust among the unemployed are often comparable to those of the employed. We also failed to find systematic associations between employment status and self-expressed confidence in the government and democratic institutions (see Annex Table 3). 19 This reiterates our earlier point that more general institutional characteristics of a society are often experienced in similar ways by different groups in the labor market. 18 Note however, that people who perceive their work to be more creative and autonomous may also tend to report higher levels of social and political activism. 19 There are small negative associations between job quality and self-expressed preferences for “strong leaders”. 22 Figure 4a. Employment Status and social outcomes The vertical axis shows the marginal probability (d-probit coefficient and 95% confidence interval) of individuals’ self-reported trust or civic engagement on being unemployed. The analysis controls for the same set of controls listed at the bottom of Table 2. Trust is based on the question, “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” Civic engagement variables are (a) whether the respondent is an active member of one or more of nine different recreational, religious, professional or political associations; (b) whether the respondent attended or would attend a demonstration; or (c) whether the respondent signed or would sign a petition. The line indicates the 95 percent confidence interval of each coefficient. If the line crosses the horizontal axis, the corresponding coefficient is not statistically significant .Note that the graphs use different scales. 23 Figure 4b Job quality and social outcomes The vertical axis shows the marginal probability (d-probit coefficient and 95% confidence interval) of individuals’ self-reported trust or civic engagement on an index assessing whether respondents think their job is cognitive, creative, or independent. Other specifications are as described in the previous Panel. Note that the graphs use different scales. VII. Discussion and conclusion There is growing awareness that uneven access to good jobs can undermine social stability in countries undergoing rapid social and economic transformations. Yet, to this point there has been surprisingly little comparative cross-country analysis how social and political experiences that relate to social cohesion vary across different groups in the labor market. In our analysis there are quite clear indications that work-related experiences of social cohesion vary between low income countries on the one hand and a growing number of high and middle income countries on the other. In lower income countries, in our sample exclusively represented by Sub-Saharan nations, conventional labor market categories, such as formal employment, appear to offer less meaningful concepts for understanding processes of social differentiation and exclusion around 24 labor market outcomes. Throughout we find only few statistically robust associations between people’s employment status (or job quality) and other indicators of social and subjective wellbeing often mentioned in the social cohesion literature. 20 We also find that subjective wellbeing is more robustly associated with average levels of income than with (formal) employment status or job quality. This suggests that, for many low income countries, concerns about the supply of jobs that guarantee at least basic incomes and job security continue to dominate over alternative preoccupations with job quality and diversity. Once one passes the threshold from low to lower middle income countries, concerns about interactions between labor market outcomes and social inequalities that are frequently raised in the context of higher income countries become more salient. Formal employment emerges as a determinant of a range of outcomes relating to social cohesion, such as the density of social associations or levels of political activism. There are also indications of an increasing association between work and life satisfaction across higher and lower middle income countries. The growing relevance of jobs for subjective and social outcomes in middle income countries raises some concerns about the future prospects for social stability in societies that undergo rapid economic transformations. Latent social tensions related to access to jobs may be aggravated as economies grow and diversify into new sectors of production. However, our cross country comparisons point to a number of statistical regularities in the link between social cohesion and economic and institutional development that may also contain lessons for how tensions associated with rapid economic growth can be mitigated. In our sample of high and low middle income countries, self-reported levels of trust and confidence in government institutions generally do not differ significantly between the employed and unemployed or across people in jobs of different quality. We also find that absolute levels of subjective wellbeing tend to increase both for people in and out of work, along with the general quality of institutions in a country. This suggests that some social and institutional features of a society, such as the general quality of public institutions and social interactions between individuals and groups, may help to mitigate conflicts within labor markets. We also find that increases in average incomes, at least up to a certain level, tend to raise levels of wellbeing for both the employed and the unemployed. This suggests that broad based growth - in particular when supported by safety nets and social protection policies that distribute the benefits of growing incomes more equally - can help to mitigate conflicts between winners and losers of economic transformation. 20 We suspect that is due to the continuously high share of workers engaged in informal income generating activities. 25 In spite of these optimistic conclusions our findings also have potentially more problematic implications for transition economies whose social and political institutions are already under stress. First, our results raise questions about the future of collective decision making in societies that have high levels of social inequality and less inclusive institutions to start with. The fact individuals who are currently out of work already participate less in social and political institutions raises concerns that future collective decision making will not involve all groups of the population. Political compromises and institutional reforms initiated to manage the social costs of economic transformation may thus not reflect the interests of all groups in society and may lead to more permanent inequalities in the access to good jobs. The second concern involves the reform of institutions that may help to mitigate work-related tensions. Social and governance institutions are difficult to change in the short term, and social protection policies and safety-nets that mitigate conflicts between winners and losers of reform may be too costly in many parts of the developing world. This complicates the challenge of managing negative consequences of economic transformation and political reform in countries with already-strained social and political institutions. Moreover, conflicts in the labor market can exacerbate existing tensions and speed-up the breakdown of peaceful interactions between individuals and groups. Finally it is difficult to predict when or why latent tensions in the labor market will erupt into actual conflict. Recent histories in the Arab World, North Africa and some high income countries suggest that even societies with relatively large differences between groups in the labor market can be remarkably stable over time, until latent tensions between groups quickly become politically ‘salient’ as economic or political outlooks change. However, whether or when these tensions actually erupt into conflict appears to be driven by a multitude of other factors that are harder to predict with the quantitative data and methods available for this study. In this sense, the broad cross-country comparisons documented in this study only represent a first step that should be followed by much more in-depth and possibly more inter- disciplinary analysis of the country-specific contexts and processes that underpin ongoing conflicts related to labor market outcomes. 26 Bibliography Alesina, A. and A. Drazan (1991). "Why are Stabilizations Delayed?" American Economic Review 81(5): 1170-1188. Altindag, T. D. and N. H. Mocan (2010). Joblessness and Perceptions about the Effectiveness of Democracy. NBER Working Paper No. 15994. Anderson, E (1999). “Code of the Street : Decency, Violence, and the Moral Life of the Inner City”, New York: W.W. Norton. Baker, J.L., Social Exclusion in Urban Uruguay. Mimeo. (2001), Washington D.C.: The World Bank. Bertaux, D. and P. Thompson (eds.) (1997). Pathways to Social Class: A Qualitative Approach to Social Mobility. Oxford: Clarendon Press. Bianchi, M. (2012). "Financial Development, Entrepreneurship, and Job Satisfaction." Review of Economics and Statistics 94(1): 273-286. Blanchflower , D. G. and R. B. Freeman (1997). "The Attitudional Legacy of Communist Labor Relations." Industrial and Labor Relations Review 50(3): 438-459. Boeri, T., A. Borsch-Supan, et al. (2001). "Would You Like to Shrink the Welfare State? Opinions of European Citizens." Economic Policy 16(32): 9-50. Botero, J. C., S. Djankov, et al. (2004). "The Regulation of Labor." Quarterly Journal of Economics 119(4): 1339-1382. Breen, R., and D. Rottman (1995). “Class Stratification: A Comparative Perspective”. Hertfordshire: Harvester Wheatsheaf. Clark, A. E. C. and A. J. Oswald (1994). "Unhappiness and Unemployment." The Economic Journal 104(424): 648-659. Coleman, J.S. (1988), 'Social Capital in the Creation of Human Capital', American Journal of Sociology, 94 (supplement), 95-120. Di Tella, R. and R. J. MacCulloch (1996). An Empirical Study of Unemployment Benefit Preferences. Working Paper N 179, IES, Oxford University. Di Tella, R., R. J. MacCulloch, et al. (2001). "Preferences over Inflation and Unemployment: Evidence from Surveys of Happiness." American Economic Review 91(1): 335-341. Di Tella, R., R. J. MacCulloch, et al. (2003). "The Macroeconomics of Happiness." Review of Economics and Statistics 85(4): 809-827. Dudwick, N. (2012). “The Relationship between Jobs and Social Cohesion: Some Examples from Ethnography.” Background paper for the WDR2013. Durlauf, S. N. (2002), Groups, Social Influences and Inequality: A Membership TheoryPerspective on Poverty Traps (mimeo). Easterlin, .R.A. (1974) "Does Economic Growth Improve the Human Lot?" in Paul A. David and Melvin W. Reder, (eds.), Nations and Households in Economic Growth: Essays in Honor of Moses Abramovitz, New York: Academic Press, Inc. 89-125. Easterlin, R. A.; McVey, A. and L. Angelescu McVey et al. (2010). "The happiness– income paradox revisited", Proceedings of the National Academy of Sciences of the United States of America107 (52): 22463–22468 Easterly, W. (2001). “The Middle Class Consensus and Economic Development”, Journal of Economic Growth 6, pp. 317-335. Easterly, W. and R. Levine (1997). "Africa's Growth Tragedy: Policies and Ethnic Divisions." Quarterly Journal of Economics 112(4): 1203-1250. 27 Easterly, W., J. Ritzen, et al. (2006). "Social Cohesion, Institutions, and Growth." Economics & Politics 18(2): 103-120. Eggers, A., C. Gaddy, et al. (2006). "Well-being and Unemployment in Russia in the 1990s: Can Society's Suffering Be Individuals' Solace?" Journal of Socio- Economics 35(2): 209–242. Fernandez, R. and D. Rodrik (1991). "Resistance to Reform: Status Quo Bias in the Presence of Individual-Specific Uncertainty " American Economic Review 81(5): 1146-1155. Flanagan, R. J. (2006). Globalization and Labor Conditions: Working Conditions and Worker Rights in a Global Economy. Oxford, Oxford University Press. Forteza, A. and M. Rama (2006). "Labor Market "Rigidity" and the Success of Economic Reforms Across more than One Hundred " Journal of Policy Reform 9(1): 75-105. Freeman, R. B. (2008). Labor Market Institutions Around the World. CEP Discussion Paper No 844. Freeman, R. B. (2009). Labor Regulations, Unions, and Social Protection in Developing Countries: Market distortions or Efficient Institutions? NBER Working Paper No. 14789. Frey, B. S. (2008). Happiness : A Revolution in Economics. Cambridge Mass., MIT Press. Giles, J., Mavridis, D, and F. Witoelar (2012). “Subjective Well-Being, Social Cohesion, and Labor Market Outcomes in Indonesia.” Background paper for the WDR 2013 Goldthorpe, J.(1992). “The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Clarendon Press. Graham, C. (2008). Measuring Quality of Life in Latin America: What Happiness Research Can (and Cannot) Contribute. Washington D.C., Inter-American Development Bank Working Paper No. 652. Graham, C. and S. Pettinato (2002). Happiness and Hardship. Opportunity and Insecurity in New Market Economies. Washington D.C. , Brookings Institutions Press. Granovetter, M. S. (1973), 'The Strength of Weak Ties', American Journal of Sociology, 78 (6), 1360-80. Gwartney, J., J. Hall, et al. (2010). Economic Freedom of the World. 2010 Annual Report, Fraser Institute. Heckman, J. J. and C. Pages (2000). The Cost of Job Security Regulation: Evidence from Latin American Labor Markets, NBER Working Paper No. 7773. Inglehart, R. and C. Welzel (2005). “Modernization, Cultural Change, and Democracy : The Human Development Sequence. New York ; Cambridge : Cambridge University Press, 2005 Kahneman, D. (1999). Objective Happiness. Well-being: the Foundations of Hedonic Psychology. D. Kahneman, E. Diener and N. Schwarz. New York, Russel Sage Foundation. Katzman, R., Seducidos y Abandonados: El Aislamiento Social de los Pobres Urbanos, Documento de Trabajo del IPES no 1 2005, Montevideo: Universidad Católica del Uruguay. Kaufmann, D., A. Kraay, et al. (2010). The Worldwide Governance Indicators: A Summary of Methodology, Data and Analytical Issues. Washington D.C., World Bank Policy Research Working Paper No. 5430. 28 Knack, S. and P. Keefer (1997). "Does Social Capital Have an Economic Payoff? A Cross-Country Investigation." Quarterly Journal of Economics 112(4): 1251- 1288. Larsen, C. A. (2007). "How Welfare Regimes Generate and Erode Social Capital." Comparative Politics 40(1): 83-101. Lindbeck, A. and D. Snower (2002). The Insider-Outsider Theory: A Survey. Bonn, IZA Working Paper No. 534. Luttmer, E. F. P. (2001). "Group Loyalty and the Taste for Redistribution." Journal of Political Economy 109(3): 500-28. Morgan, S. L., Grusky, D. B., and G. S. Fields, (eds.) (2006), Mobility and Inequality. Frontiers of Research in Sociology and Economics (Stanford Ca.: Stanford University Press). Norton, A., and A. de Haan. 2012. “Social Cohesion: Theoretical Debates and Practical Applications with Respect to Jobs.” Background paper for the WDR 2013. OECD (2011). Perspectives on Global Development 2012: Social Cohesion in a Shifting World. Paris, OECD. Olson, M. (1971). “The Logic of Collective Action : Public Goods and the Theory of Groups”. Cambridge, Mass. Harvard University Press. Olson, M. (1982). “The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities”. New Haven. Yale University Press Petesch, Pi. ( 2012). “The Exponential Clash of Conflict, Good Jobs, and Changing Gender Norms in Four Economies.” Background paper for the WDR 2013. Rama, M. and Artecona (2002). A Database of Labor Indicators Across Countries The World Bank, Washington D.C. Rodrik, D. (1997). Has Globalization Gone too Far ? Washington D.C. Rodrik, D. (1999). "Where Did All the Growth Go? External Shocks, Social Conflict, and Growth Collapses." Journal of Economic Growth 4(4): 385–412. Rothstein, B. and E. M. Uslaner (2005). "All for All: Equality, Corruption, and Social Trust." World Politics 58(1): 41-72. Rueda, D. (2005). "Insider–Outsider Politics in Industrialized Democracies: The Challenge to Social Democratic Parties." American Political Science Review 99(1): 61-74. Rueda, D. (2006). "Social Democracy and Active Labor-Market Policies: Insiders, Outsiders and the Politics of Employment Promotion." British Journal of Political Science 36: 385-406. Schneider, F. (2004). The Size of the Shadow Economies of 145 Countries all over the World: First Results over the Period 1999-2003. Bonn, IZA Working Paper No. 1431. Schneider, F. and D. H. Enste (2000). "Shadow Economies: Size, Causes, and Consequences." Journal of Economic Literature 38: 77-114. Shayo, M. (2009). "A Model of Social Identity with an Application to Political Economy: Nation, Class, and Redistribution." American Political Science Review 103(2): 147-174. Shiller, R. (1996). Why Do People Dislike Inflation? NBER Working Paper No. 5539. Wietzke, F.B. (2009). "Neighbourhood Effects: A Critical Review of the Literature". Background paper for the World Bank's World Development Companion Report on spatial inequality in Latin America. 29 Winkelmann, L. and R. Winkelmann (1998). "Why Are the Unemployed So Unhappy? Evidence from Panel Data." Economica 65(1-15). Woolcock, M. (2011). “What Distinctive Contribution Can Social Cohesion Make to Development Theory, Research, and Policy?” Paper presented at the Organisation for Economic Cooperation and Development International Conference on Social Cohesion and Development, Paris, January 20. 30 Annex Table 1. Descriptive statistics. WVS Round V High income High middle income Lower middle income Low income Total mean sd mean sd mean sd mean sd mean sd life satisfaction 7.45 1.72 7.07 2.26 5.86 2.51 5.46 2.26 6.74 2.31 lower class 0.04 0.19 0.13 0.34 0.17 0.38 0.35 0.48 0.13 0.34 upper class 0.29 0.45 0.19 0.39 0.16 0.37 0.14 0.35 0.19 0.39 trust 0.43 0.50 0.21 0.41 0.24 0.42 0.16 0.37 0.21 0.41 demonstration 0.70 0.46 0.50 0.50 0.44 0.50 0.59 0.49 0.50 0.50 petition 0.89 0.31 0.54 0.50 0.41 0.49 0.59 0.49 0.54 0.50 associations 0.94 1.21 0.67 1.19 0.86 1.42 1.14 1.62 0.67 1.19 Confidence in government 0.36 0.48 0.49 0.50 0.45 0.50 0.32 0.47 0.49 0.50 preference for strong leader 0.21 0.41 0.40 0.49 0.31 0.46 0.31 0.46 0.40 0.49 unemployed 0.09 0.29 0.15 0.35 0.19 0.39 0.21 0.41 0.15 0.36 job quality 6.01 2.09 5.27 2.13 4.60 2.11 4.97 1.79 5.32 2.15 self employed 0.10 0.31 0.18 0.38 0.22 0.42 0.36 0.48 0.19 0.39 part time 0.14 0.35 0.10 0.30 0.12 0.32 0.07 0.25 0.11 0.31 unskilled 0.05 0.21 0.14 0.35 0.17 0.37 0.17 0.38 0.13 0.33 farmer 0.01 0.11 0.08 0.26 0.07 0.25 0.15 0.36 0.07 0.25 work very important 0.52 0.50 0.68 0.47 0.80 0.40 0.81 0.39 0.68 0.47 work important 0.42 0.49 0.27 0.44 0.17 0.37 0.17 0.38 0.27 0.45 age 42.29 12.92 38.31 13.04 37.42 12.78 35.39 12.89 38.82 13.11 age2 1954.98 1147.15 1637.48 1110.65 1563.33 1085.18 1418.80 1108.03 1678.91 1126.48 income 5.42 2.54 4.75 2.34 4.38 2.10 4.37 2.15 4.79 2.35 savings 0.30 0.46 0.21 0.41 0.22 0.41 0.28 0.45 0.24 0.43 widowed or divorced 0.12 0.32 0.08 0.27 0.08 0.27 0.08 0.27 0.09 0.28 31 married 0.65 0.48 0.64 0.48 0.67 0.47 0.61 0.49 0.65 0.48 female 0.49 0.50 0.43 0.49 0.37 0.48 0.39 0.49 0.43 0.49 debt 0.07 0.26 0.13 0.33 0.13 0.34 0.07 0.26 0.11 0.31 religious 0.13 0.34 0.19 0.39 0.21 0.41 0.34 0.47 0.19 0.39 secondary education 0.50 0.50 0.44 0.50 0.40 0.49 0.23 0.42 0.43 0.49 higher education 0.21 0.41 0.16 0.36 0.15 0.36 0.09 0.29 0.16 0.37 children 1.51 1.32 1.72 1.73 2.07 1.91 2.16 2.28 1.80 1.76 GNI pc 25283.22 9201.99 4539.87 5501.26 949.90 497.53 213.98 48.51 9417.53 11771.39 labor law index 0.50 0.23 0.44 0.17 0.45 0.20 0.53 0.11 0.47 0.20 rule of law index 1.55 0.37 -0.15 0.61 -0.52 0.55 -0.78 0.28 0.13 0.99 gov' effectiveness 1.67 0.42 0.13 0.54 -0.55 0.50 -0.96 0.37 0.25 1.02 unemployment rate 6.94 3.13 10.51 6.75 10.28 4.97 5.69 2.45 9.21 5.66 gov' transfers 5.61 1.48 7.85 1.12 8.18 1.36 9.15 0.88 7.45 1.71 32 Table 2. effect of national institutional attributes on the wellbeing of the employed and unemployed Panel A Panel B Life satisfaction of the unemployed Life satisfaction of the employed (1) (3) (5) (7) (9) (11) Rule of law (‘96- 0.555*** 0.274*** ‘05) (0.104) (0.028) Gov’ effectivenss 0.957*** 0.783*** (’96-‘05) (0.075) (0.026) Labor laws index 1.372*** 0.782*** (Botero et al. 2004) (0.294) (0.093) sigma 2.629*** 2.598*** 2.669*** 2.270*** 2.241*** 2.207*** (0.029) (0.029) (0.033) (0.012) (0.011) (0.012) Constant 5.025*** 5.289*** 4.381*** 6.217*** 6.104*** 5.711*** (0.396) (0.390) (0.545) (0.168) (0.166) (0.204) Observations 5393 5393 4178 30177 30177 26209 Estimates based on the full sample of the World Value Survey Wave 5. The estimation sample in Panel includes only unemployed individuals and the sample in Panel B only those currently in work. The dependent variable is self-reported life satisfaction measured on a 10pt scale, with higher values denoting higher life satisfaction. Coefficients are calculated with a tobit model to account for possible over reporting on the highest score of the scale. All estimates include controls for GNI, share of government transfers in the Budget (from Gwartney, J., J. Hall, et al. 2010), as well as the same set of personal controls as in Table 2. *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. 33 Table 3. Job Status and political attitudes. World Value Survey High income Higher middle income Low income Lower middle income Dependent Confidence in Confidence in Confidence in Confidence in Confidence in Confidence in Confidence in Confidence in variable government government government government government government government government unemp -0.007 0.018 -0.025 -0.005 (0.018) (0.012) (0.024) (0.015) Job qual -0.000 0.001 -0.002 -0.001 (0.003) (0.002) (0.007) (0.003) Observations 11827 9614 17534 13481 3267 2090 10981 6838 Dependent Preference for Preference for Preference for Preference for Preference for Preference for Preference for Preference for variable strong leader strong leader strong leader strong leader strong leader strong leader strong leader strong leader Unemp 0.023 0.012 -0.041* 0.029** (0.014) (0.012) (0.023) (0.013) job qual -0.009*** -0.006** 0.011* -0.007** (0.002) (0.003) (0.007) (0.003) Observations 11827 9614 16658 12742 3267 2090 12277 7886 All estimates control for age, age squared, family status (married, widowed, or divorced), level of education (secondary or higher), number of children, gender, religious and work attitudes and income. Source: authors’ estimates based on WVSs round V. Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. 34