The Economic Implications of Social Cohesion [1 ed.] 9781442681149, 9780802037367

Citation preview

THE ECONOMIC IMPLICATIONS OF SOCIAL COHESION

Studies in Comparative Political Economy and Public Policy Editors: Michael Hewlett, David Laycock, Stephen McBride, Simon Fraser University. Studies in Comparative Political Economy and Public Policy is designed to showcase innovative approaches to political economy and public policy from a comparative perspective. While originating in Canada, the series will provide attractive offerings to a wide international audience, featuring studies with local, subnational, cross-national, and international empirical bases and theoretical frameworks. Editorial Advisory Board Colin Bennett, University of Victoria William Carroll, University of Victoria William Coleman, McMaster University Barry Eichengreen, University of California (Berkeley) Jane Jenson, Universite de Montreal Rianne Mahon, Carleton University Lars Osberg, Dalhousie University Jamie Peck, Manchester University John Ravenhill, University of Edinburgh Robert Russell, University of Saskatchewan Grace Skogstad, University of Toronto Rand Smith, Lake Forest College Kent Weaver, Brookings Institution For a list of books published in the series, see p. 251

EDITED BY LARS OSBERG

The Economic Implications of Social Cohesion

U N I V E R S I T Y OF TORONTO PRESS Toronto Buffalo London

www.utppublishing.com University of Toronto Press Incorporated 2003 Toronto Buffalo London Printed in Canada ISBN 0-8020-3736-4 (cloth)

Printed on acid-free paper

National Library of Canada Cataloguing in Publication Data The economic implications of social cohesion / edited by Lars Osberg. (Studies in comparative political economy and public policy) Includes bibliographical references ISBN 0-8020-3736-4 1. Economic development - Social aspects. 1970-. I. Osberg, Lars II. Series. HM548.E26 2003

338.9

2. Social history -

C2002-903318-7

University of Toronto Press acknowledges the financial assistance to its publishing program of the Canada Council for the Arts and the Ontario Arts Council. University of Toronto Press acknowledges the financial support for its publishing activities of the Government of Canada through the Book Publishing Industry Development Program (BPIDP).

Contents

Introduction — LARS OSBERG

3

Do Borders Matter for Social Capital? Economic Growth and Civic Culture in U.S. States and Canadian Provinces 19

JOHN F. HELLIWELL

Social Capital, Social Cohesion, Community: A Microeconomic Analysis JEFF DAYTON-JOHNSON

43

Social Cohesion and the Well-being of Canadian Children SHELLEY PHIPPS

79

Social Cohesion and Health JOHN N. LA VIS AND GREGORY L. STOOD ART

121

Social Cohesion and Voluntary Activity: Making Connections FRANCES WOOLLEY

150

Communities and Economic Prosperity: Exploring the Links JANE FRIESEN

183

Social Cohesion and Macroeconomic Performance M.C. McCRACKEN

213

Many Happy Returns: How Social Cohesion Attracts Investment DICK STANLEY AND SANDRA SMELTZER

Notes on Contributors

247

231

This page intentionally left blank

THE ECONOMIC IMPLICATIONS OF SOCIAL COHESION

This page intentionally left blank

Introduction LARS OSBERG

In day-to-day life, most people are well aware of the advantages of cooperation. Soccer teams that pass well will almost always dominate their opponents, if the other team are all individualistic glory hounds who hog the ball to maximize their personal goal scoring. Families that can agree on what to do on holiday trips have a better time if some members do not sulk all the way to their destination. Firms where workers and managers cooperate tend to be more productive and profitable than those mired in conflict. In the practical reality of everyday living, there are abundant illustrations of the benefits of cooperating for a common goal. That cooperation can be called 'team behaviour' - and in recent years a growing literature has argued that such behaviour pays off for societies, as well. Why then might societies differ in their propensity to cooperate? What are the pay-offs to such cooperation? If one looks around the world, it is clear that there are big differences between countries in the level of civil strife and conflict. In some regions, there is a long history of violence, sometimes between ethnic groups and sometimes derived from political differences, which seems extremely difficult to escape once it becomes established. Other societies may avoid civil war, but still have fractious bureaucracies and combative industrial relations that make daily business a constant succession of unnecessary difficulties. Once expectations of other people's uncooperativeness become entrenched, they seem hard to eliminate, but how can one explain why some countries have much more conflict than others? One answer is that societies differ in their social cohesion and that more cohesive societies find it easier to agree on, and to implement, social decisions. Of course, this sort of answer just provokes the questions: What is social cohesion? Why might social cohesion matter?

4

Lars Osberg

This volume's objective is to make a start on the answers. It focuses particularly on the economic implications of social cohesion. If societies with more social cohesion are able to produce more, that additional output will make it easier to resolve distributional conflicts, and a 'virtuous circle' or positive feedback loop of more cohesion / more cooperation / more output / more cohesion may become established. Conversely, the economic costs to less social cohesion and less cooperation will mean there is less total output to share, which will tend to heighten distributional conflict and may set up a 'vicious circle' in which lower output reinforces lower cohesion. The contributors to this volume begin, therefore, with the supposition that the social framework that conditions economic decision making may sometimes be crucial to those decisions. If so, the research agenda is to spell out why that might be the case, how one might test the relationship between social cohesion and economic outcomes, and what difference it might make to economic and social policy. This book does not claim to have completed this research agenda - the scope of interactions is far too vast, and the field is developing far too quickly for that to be feasible. However, this volume does examine the impact of social cohesion on a number of important economic issues, including health, the well-being of children, macroeconomic performance, voluntary activity, the role of community institutions, investment, and regional development. By tracing the connections between these economic outcomes and their social framework, the book hopes to contribute to a broader understanding of the interaction between economic processes and their social context. The authors in this volume start from a common background in economics, but they address a set of issues that economists have typically tended to leave to 'somebody else.' As a discipline, economics has made a great deal of intellectual progress by relying (most of the time) on the assumption of ceteris paribus - that one can analyse economic processes holding 'other things constant.' By assuming that the social framework for economic decision making remains unchanged (or at least changes slowly, and in response to non-economic influences), economists have been able to focus their attention. There is plenty of complexity in the incentives, and the constraints, that influence the decisions individuals and firms make within particular markets, and economists have made great progress in their analysis. For many issues of great social importance within countries, 'other things constant' is a plausible assumption - which is often of great convenience to the analysis of complex market processes. Hence, omission of attention to social context is sometimes justifiable.

Introduction 5 All the same, this book argues that there are reasons to think that sometimes the social framework of economic processes matters crucially and changes endogenously. This will not exactly be news for many non-economist readers, and some of the greatest economic thinkers have also emphasized the importance of social context. For example, in development economics Simon Kuznets argued that the distinguishing characteristic of the modern economic epoch is 'the extended application of science to problems of economic production' (1966: 9). However, Kuznets emphasized that this could not happen in the absence of 'a proper climate of human opinion in which both the pursuit and use of science could be fostered,' and he argued that 'the broad views associated with the modern economic epoch can be suggested by three terms: secularism, egalitarianism and nationalism' (1966: 12). Kuznets was an optimist, who thought that modern economic growth had a fairly long course to run, even in the most developed countries.1 However, among those concerned with the social context of economics there is also a pessimistic tradition which argues: 'Capitalism tends to destroy itself ... [because] ... capitalist activity, being essentially rational, tends to spread rational habits of mind and to destroy those loyalties and those habits of super- and subordination that are nevertheless essential for the efficient working of the institutionalized leadership of the producing plant: no social system can work which is based exclusively upon a network of free contracts between (legally) equal contracting parties and in which everyone is supposed to be guided by nothing except his own [short run] utilitarian ends' (Schumpeter 1950: 417). Schumpeter's analysis should be seen in the context of a long-standing concern with the social framework of market processes and the social instability entailed by pursuit of purely individual advantage. In the mid-nineteenth century, for example, Alexis de Tocqueville tried to explain to a European audience the vibrancy of American economic and political life, and he argued that 'the Americans Combat Individualism by the Principle of Interest Rightly Understood.' De Tocqueville claimed that Americans 'show with complacency how an enlightened regard for themselves constantly prompts them to assist each other, and inclines them willingly to sacrifice a portion of their time and property to the welfare of the State' (1961: 146). Even earlier, Adam Smith had noted in his book Theory of Moral Sentiments2 (1986: 110-12): The regard to those general rules of conduct, is what is generally called a sense of duty, a principle of the greatest consequence in human life, and the only principle by which the bulk of mankind are capable of directing their actions ... Upon the tolerable observance of these duties depends the very existence of human society, which would crumble into

6

Lars Osberg

nothing if mankind were not generally impressed with a reverence for these important rules of conduct.' Evidently, concerns about the importance of the social and ethical framework of economic life have been around for some time. Thus, the development of a new literature that discusses social capital and social cohesion can be seen partly as an attempt to be more precise about some very old hypotheses in social analysis. Neither term - 'social capital' or 'social cohesion' fits the normal economics mould. Economics is a discipline that prides itself on precision, but both social capital and social cohesion are hard to define, and often are confused with each other. Economists usually start from the perspective of a selfish, utility-maximizing individual, whose interaction with others is limited to buying and selling in the marketplace. Social capital and social cohesion, however, are both about social relationships, group identities, and thus the non-market dimensions of life. Nonetheless, the growth in concern for social capital and social cohesion is unmistakable.3 Some of the reason for this may lie in recent political events. Since 1990, the demise of the Soviet regime in Eastern Europe and the former USSR has provided a dramatic example of what can happen when ceteris is not paribus, and the social framework of economic processes changes rapidly and substantially. Undoubtedly, terms like 'social capital' and 'social cohesion' have become hot in recent years partly because the vocabulary of economics needed expansion if the issues of the transition economies were to be adequately addressed. When the Berlin Wall fell in 1989, there was a great deal of optimism among economists for the economic future of Eastern Europe. Although, in retrospect, that optimism makes embarrassing reading, at the time most of the world's economists thought that economic growth in the USSR and Eastern Europe would be rapid in the post-Soviet era. Because these nations had technically sophisticated, highly educated labour forces and a great deal of capital, many analysts expected that the elimination of the dead hand of communist central planning would unleash pent-up potential for rapid growth. These expectations were based on the simple perspective that economic production occurs when capital, labour, and human capital are combined at the workplace. Since many economists thought, and continue to think, that the price signals of an unregulated market are the most effective way possible of coordinating economic activities, they concluded that as soon as Eastern Europe and the USSR acquired a market system, good things would happen. If this was all there was to it, history would have turned out differently. During the 1990s, it was profoundly shocking to many professional economists to see the decline in living standards that has actually occurred in post-

Introduction

7

Soviet nations, together with the rise of 'gangster capitalism' in much of the old Soviet bloc. This historical experience has pushed many economists to ask, What went wrong? It has also prompted a resurgent recognition of the importance of the social context of market processes. In addition, during the 1990s, some domestic trends in Western nations amplified concerns about the social framework. Many Western nations have faced social strains in accommodating migration flows that are increasingly diverse, both ethnically and racially, and there has been disturbing evidence of increasing political polarization in Europe and the rise of xenophobic political tendencies. In North America, there has also long been widespread concern over the implications of fragmenting families and deteriorating urban schools and neighbourhoods. One of the earliest uses of the term 'social capital' was in James Coleman's 1988 discussion of the determinants of neighbourhood differences in school attainment.4 Putnam (1993) emphasized the importance of social capital for good government and strong economic growth in Italian regions. His more recent (2000) discussion of the long-run decline in associational life in the United States reinforces the theme of underinvestment in social capital and argues strongly that there are large long-term costs to this trend. Social capital and social cohesion may therefore be new terms, and recent events may have given a boost to their popularity. These are not, however, new concepts in social science. There has long been a concern in Western nations with the issues raised by the social capital and social cohesion literature, even if early writings tended to be broader in focus, and less quantitative in orientation, than is favoured by the modern social science tradition. Hence, a number of powerful arguments motivate what the Organization for Economic Cooperation and Development (OECD 2001: 39) calls 'a growing awareness in the economic literature of the importance of social networks and trust in supporting collective endeavors.'5 Thus far this introduction has not attempted a definition of either social capital or social cohesion. As the OECD notes, social capital 'provides a useful umbrella term for those aspects of society which, although difficult to measure and incorporate into formal models, are widely thought to be an important determinant of long run economic success. For some economists (not all) the intuition that "society matters" is strong enough to outweigh the current absence of much in the way of theoretical underpinning' (OECD 2001:39). The aim of this book is to fill some of that theoretical void. However, we face the problem that the literature is currently in transition and a number of definitions of key terms are in use concurrently. Debate continues between

8 Lars Osberg those who favour narrow and broad use of terms. The OECD offers one example of a 'lean' definition of social capital as 'networks together with shared norms, values and understandings that facilitate cooperation within or among groups' (2001: 41). In their chapter in this volume, Dick Stanley and Sandra Smeltzer build on Stanley's earlier work (1999), where he defines social cohesion in a way somewhat similar to that offered by the OECD for social capital. Social cohesion, according to Stanley, is 'the bonding effect of society which arises spontaneously out of the unforced willingness of individuals to enter into relationships with each other and to contribute to the collectivity in order to enhance their survival, sustenance and success.' Evidently, there is a substantial overlap between the two concepts. As is made apparent in this volume, although it is sometimes useful to emphasize areas of commonality and overlap in definitions, it is also sometimes useful to concentrate attention on the differences. Although various definitions serve as a starting point, this volume does not enforce a single overarching definition or set of definitions. The editor's opinion is that it would be premature to do so at this stage of development of the literature on social cohesion and social capital. Thus, the contributors to this volume have used the conceptualization that makes most sense in the context of their specific work. Contributors were invited to focus on particular issues - health implications, the well-being of children, the voluntary sector, the role of community institutions, investment, macroeconomic outcomes, and regional development. The general philosophy guiding the use of terms can be summarized as 'Get the right tool for the job.' It was anticipated that the definition of social cohesion most appropriate for an analysis of, for example, the well-being of children may differ from the use of terms most appropriate for an analysis of health or voluntary sector issues. Authors were asked to be clear about their own use of terms, but they were not required to all adopt exactly the same use of terms. In practice, however, there is substantial overlap. This pragmatic flexibility in terminological usage is fairly common in economics (as well as in other fields). As an example, one can mention the term 'human capital.' By now the term 'human capital' has passed into common useage and proved useful for much work on labour market analysis, even though its precise definition differs with different authors. Originally Becker (1964: 1) thought of human capital as 'activities that influence future monetary and psychic income by increasing the resources in people.' This conceptualization has proven useful for, among other things, the study of the educational, training, and mobility decision making of both individuals and firms. Nevertheless, the Becker definition is not quite the same as that favoured

Introduction 9 by the OECD, for which human capital is 'the knowledge, skills, competencies and attributes embodied in individuals that facilitate the creation of personal, economic and social well-being' (2001: 18). The OECD definition includes attributes of individuals that are innate, fixed, and passively acquired (like native intelligence or personal beauty), whereas the Becker perspective emphasizes the current choices that influence future returns. In addition, the OECD definition does not distinguish between individual attributes such as race or gender that may privilege some individuals at the expense of others6 and characteristics that are socially productive (such as education), while Becker's use of the term 'resources' can be read as meaning only attributes that increase personal productivity. Perhaps because of the underlying ambiguities in the concept, labour economics texts often adopt the strategy of avoiding any explicit definition of 'human capital' and confine themselves to just listing some of the things that qualify.7 The frequent avoidance of explicit definition and the ambiguities in current definitions do not imply that 'human capital' is useless as a concept for analysis. Similarly, 'social cohesion' can be a useful concept for analysis, even while some authors use slightly differing definitions of it - as long as they are clear about their definitions. It seems most reasonable to argue that one should ask both the social cohesion and human capital literatures for a similar level of (non)uniformity in terminological usage, and accept that, in practice, it will often be of value to have a slightly different usage of terms to accommodate different empirical issues. The function of an introductory chapter is both to outline the chapters to come and to indicate the differences and common elements in the definition and use of terms. John Helliwell, for example, uses the terms 'social capital' and 'civic culture' interchangeably. The volume begins with Helliwell's analysis of economic growth and civic culture in U.S. states and Canadian provinces. In focusing on the World Values Survey (WVS) measure of trust as an index of the level of social capital, and in emphasizing the role of voluntary associational life, Helliwell's work is closely linked to the original framework of Robert Putnam (1993). As Helliwell notes, both Canada and the United States are far from homogeneous internally with respect to measured trust and associational life, and there are complex historical origins of the differences among Canadian provinces and U.S. states. In the social cohesion literature, the 1997 article by Knack and Keefer on the economic pay-off to social capital has been much cited; for example, see the Stanley and McCracken chapters in this volume. Knack and Keefer's main finding is a higher rate of economic growth in countries with higher levels of social capital. The question remains, however, whether this effect is

10 Lars Osberg specific to particular phases of capitalist economic development and whether the same impact of social capital can be expected within affluent countries. Because the growth rates of U.S. states and Canadian provinces have been dominated by the conditional convergence of output levels, Helliwell finds it difficult to detect an added influence of trust. Nevertheless, there is some evidence that in the United States internal migration tends to favour states with higher perceived levels of trust. Although Helliwell does not distinguish particularly between social capital and social cohesion, Jeff Dayton-Johnson does in his chapter. Dayton-Johnson's concern is to lay out explicitly the theoretical microeconomic foundations of social capital, social cohesion, and community and then to distinguish carefully between these concepts. He argues that in many ways measurement has gotten ahead of theory in the literature on social capital and social cohesion, and he therefore uses a game theoretic framework to develop the implications of these concepts. In Day ton-Johnson's view, it is useful to think of social capital as an individual asset, similar to human capital, that requires individual agents to set aside current resources in the hope of a future return. The idea underlying social capital in this sense is that individuals invest in social relationships and reputations and by doing so receive a claim on the returns to mutual cooperation within those relationships. Day ton-Johnson thinks of social cohesion as a society-level characteristic and defines it as an environmental or state variable that changes over time and adheres in the population of players. Social cohesion is not an individual asset, and its level is not subject to the decision of any particular person. However, if individuals play repeated games with an ever-changing population of other agents, they will want to know the aggregate stock of social capital investments made by all agents in the past because their current decision to cooperate (and thereby increase the future stock of aggregate social capital) or to defect (and thereby diminish social cohesion) will depend on this aggregate state variable. Day ton-Johnson distinguishes the idea of 'community' from the terms 'social capital' and 'social cohesion' by noting that a community exists if there is a structure of information and a social standard of behaviour that supports positive amounts of cooperation. This idea of 'shared values and communities of interpretation' is a stronger condition than either individualistic calculation of the pay-off to individual reputation or observation of the aggregate past cooperative behaviour of other individuals. Day ton-Johnson goes on to analyse the implications of social capital investment, social cohesion, and community in segmented societies. It is clear that the class of models that his analysis produces has multiple equilibria, some of which are clearly preferable to others. There is no guarantee that the cooperation essential for pros-

Introduction

11

perity will emerge spontaneously, and there is a potential role for state action to move to a preferable equilibrium. The chapter by John Lavis and Greg Stoddart examines the relationship between social cohesion and health, defining social cohesion as 'the networks, norms, and trust that bring people together to take action.' This definition is pretty close to that of the OECD and is consistent with DaytonJohnson's. Lavis and Stoddart are particularly concerned with disentangling the pathways and mechanisms by which social cohesion may affect health. If Canadians are asked what distinguishes the role of the state in Canadian and U.S. society, they are likely to think of medicare. If asked whether this is good public policy, they may mention the greater life expectancy of Canadians. But what causes what? Do Canadians have medicare because Canada has more social cohesion and have longer average life expectancy because of the greater availability of medical care? Or does social cohesion contribute directly to health? Lavis and Stoddart explore both the direct and indirect effects of social cohesion on health status. Because individuals who are embedded in dense networks of social relationships tend both to be healthier and to receive more informal care, social cohesion may directly affect the health of individuals. Because cohesive societies can organize themselves more effectively, for purposes such as public health or workplace safety, social cohesion may also indirectly affect the health of a population overall by enabling collective action on issues that affect health. However, it is also possible that social cohesion and health owe their positive correlation to a common cause - it can plausibly be argued that low levels of economic inequality,8 or high levels of government performance, contribute to both social cohesion and population health. Lavis and Stoddart use the World Values Survey to analyse the relationship between self-reported health status, trust, and voluntary associational membership. They also survey the literature on the social determinants of health and conclude that there is abundant evidence to show that social cohesion can have health consequences. Empirical research suggests that the health consequences of the level of social cohesion may be more significant than the average policy maker or citizen now believes. Like most of the literature, the work of Lavis and Stoddart studies outcomes among adults. Shelley Phipps finds that for children the connection between social cohesion and well-being may be different than for adults. Phipps suggests that it is useful to think of social cohesion in terms of a definition provided by Stanley (1997: 1): 'Social cohesion is about all of us "sticking together": it is the bonding effect of that web of social relationships through which individuals are attached to each other in a society, and through

12 Lars Osberg which they help each other, knowingly or inadvertently, to achieve their full potential.' To analyse effectively the connection between social cohesion and the well-being of children, one must address the ways in which children differ from adults and how those differences influence the relevant definitions of terms. For adults, it may be reasonable to think of individual well-being as determined by the choices made by rational utility-maximizing persons with fixed preferences. However, 'growing up' is a defining characteristic of childhood; that is, for children, both tastes and the ability to calculate one's own best interest are changing substantially over time. As well, the general definition of social cohesion as 'the bonding effect of that web of social relationships through which individuals are attached to each other in a society' does not fully capture a child's world. Compared with adults, children have a limited (though expanding) conception of the wider world outside the household. Phipps argues for a child-centred perspective, in which children's wellbeing is best understood as a web of mutually interdependent functionings. She estimates the empirical importance of childhood social relationships both for the current well-being of the child and children's eventual adult outcomes. The links between (1) parental and neighbourhood social supports and changes in family residence and (2) current child outcomes indicate that social cohesion variables have relatively 'big' effects on child outcomes compared, for example, with the effects of poverty or single-parent family status. If one is to cut through the verbiage that sometimes overwhelms social science discourse, a real world example is often useful. Children's soccer leagues are a nice, concrete example of social cohesion in action and of the creation of social capital. The viability of children's sports leagues depends on social capital, as expressed in parental involvement to ferry players about and in the volunteer coaches and managers who run practices and games. The bonding effect of team play is a concrete example of small-scale social cohesion and the positive feedbacks are clear: children learn cooperative behaviour and the advantages of working together as a team on the field, and teams that cooperate well win more often. As well, participants retain a certain amount of fitness and have, during their teenage years, a socially acceptable substitute for gangs and a bounded outlet for aggression. At the same time, parents are building their own social networks in the stands. However, it can easily be forgotten that none of this could happen without the actual soccer fields. In Canada, it is the municipal public sector that usually provides this essential infrastructure, and for Frances Woolley, the central question for public policy is whether in general the public sector complements or competes with the voluntary activity that brings stronger social cohesion. Woolley defines

Introduction

13

social cohesion as 'the ongoing process of developing a community of shared values, shared challenges, and equal opportunity within Canada, based on a sense of trust, hope, and reciprocity among all Canadians.' Her chapter provides an overview of theoretical and cross-national evidence on the connection between social cohesion and voluntary activity. Woolley surveys economic theories of voluntary activity and finds little evidence that reductions in government spending lead to greater voluntary activity. Rather, shared values, the absence of extreme income inequality, and shared tastes are necessary for collective action whether through governments or voluntary organizations. Income inequality and social fragmentation erode the basis for voluntarism, just as they put stress on 1960s and 1970s models of government. Woolley uses recent Canadian evidence to examine the link between voluntary activity and trust. The relationship between voluntarism and other social cohesion indicators, such as shared values or tolerance, is complex. Only a minority of voluntary activity appears to create bridging social networks, reaching those at risk of social exclusion. Instead, voluntarism appears to arise from the embeddedness of individuals in relationships. Employment, marriage, having children, and education are part of a virtuous cycle in which voluntarism reinforces, and is reinforced by, existing social networks. As a consequence, there are distinct limits to the replacement of government with voluntary service delivery. Arguably, what volunteers do best is provide highquality, personal goods in relatively small quantities. More general reliance on volunteer delivery can lead to conflict between accountability and volunteer motivation, discrimination in the provision of services, inadequate supply, and increased income inequality. Jane Friesen sounds a similarly cautious note. She uses the term 'community,' rather than 'social cohesion,' defining community as 'a group of people who are likely to experience repeated direct face-to-face social contact because of some characteristics they have in common.' Friesen notes the twosided potential of dense networks of social interaction for both positive and negative outcomes. The social norms and values perpetuated in stable communities may carry benefits in the form of norms of reciprocity and trust which engender low transaction costs and thereby assist prosperity, or they may reinforce attitudes of social distance and intolerance and cause economic losses and unrealized gains from trade. In principle, social cohesion can bond people into small subgroups within the broader society, or it can bridge the differences between individuals in different social groups. One reason why the distinction between bridging and bonding may matter is if individuals are locked into local communities in a cycle of advantage and

14 Lars Osberg disadvantage. Friesen is interested in the impact of communities on educational attainment, much as Coleman was in the United States. She uses Canadian census data to document the relative homogeneity of income within neighbourhoods and its increasing heterogeneity across neighbourhoods. Friesen's conclusion is that the extent to which community life is able to generate spillovers between different income groups is very limited, and in most Canadian cities it is declining. The quality of local public schools can be an important determinant of increasing residential segregation. Canadian society relies heavily on the public school system to socialize youth to a common understanding of shared values and communities of interpretation. Friesen analyses the role that neighbourhood characteristics play in determining test scores in high schools in British Columbia. She argues that social trends such as increasing segregation of families by income, growing numbers of single-parent families (many of whom have low incomes), the greater time stress of dual-earner households, and high levels of immigration affect the ability of the school systems to produce graduates able to communicate effectively, express themselves in a nuanced way, and grasp complex social ideas. These tendencies may make it more difficult for Canadian society to remain cohesive in the future. Furthermore, Friesen's discussion of the two-edged nature of social cohesion reminds us that the public policy objective is to develop communities that are socially cohesive in a positive way. For both Jane Friesen and Mike McCracken, social interaction and the formation of the norms and values that support social cohesion are central issues. McCracken starts from a very broad view of the idea of social capital as 'family and kinship connections; wider social networks, or "associational life" - networks of civic engagement; cross-sectional linkages, or contacts spanning differences in sector or power; political capital - the informal institutional arrangements for mediating conflict; institutional and policy framework - the set of formal rules and norms (constitutions, laws, regulations, and policies) that regulate public life in society; and social norms and values.' McCracken's particular emphasis, however, is on the networks to which individuals belong and on the costs to individuals of being socially excluded from network membership. Time-use diaries indicate that most people actually spend a very small part of their day in civic or voluntary activities. Paid work (for those employed full-time) is the dominant networking or socializing activity outside the family. As a consequence, exclusion from paid employment is the most important determinant of social capital. McCracken notes that the thrust of much recent economic policy is to exclude individuals from socially valuable networks,

Introduction

15

and he argues for institutional change in the way future economic policy is formulated. Dick Stanley and Sandra Smeltzer's focus is similarly macroeconomic in perspective, because they are interested in the influence of social cohesion on economic growth and investment at the macro level. Stanley and Smeltzer argue that social cohesion reduces expected transaction costs through the provision of an amenity 'bonus,' through reducing the need for defensive necessities and through the reduction in potentially costly political and labour instability. Social cohesion reduces expected costs to firms by increasing worker productivity through a reduction in the incidence of individual social dysfunction, increased employee satisfaction, and the increased ease with which new ideas are disseminated. Moreover, the evolution of the nature of work in our knowledge-based society is likely to make social cohesion even more important in maintaining a favourable investment climate in the future than it now is. Although the contributors to this volume employ slightly different definitions of the term 'social cohesion,' there are common denominators. The idea of 'networks, norms, and institutions that facilitate collective action' runs through most of the chapters in this volume; the nature of that collective action remains a separable issue. There is a common recognition of the complexity of implications of social capital and social cohesion. Stanley is the most positive about the potential implications of higher levels of social cohesion, but other contributors (especially Friesen) lay great emphasis on the ambiguity of social cohesion and its potential to support either negative or positive types of collective action. As Lavis and Stoddart point out, disentangling cause and correlation in the analysis of social cohesion is far from straightforward. Woolley paints a nuanced picture of the relationship between social cohesion, voluntarism, and the state. Phipps focuses on the different measures appropriate to adults and children. McCracken emphasizes the reciprocal impacts between institutions and good economic performance, and Helliwell stresses the heterogeneity of civic cultures within nations (i.e., within Canada and the United States) and the ambiguity of a link to economic growth. This book suggests some of the pathways by which social cohesion can affect particular economic outcomes and several of the chapters offer empirical evidence consistent with those hypotheses. We hope that it serves to give some greater specificity to the relationships between social cohesion and economic outcomes. However, we are not at the point where we can estimate the magnitude of these implications with much confidence. Hence, to paraphrase (and amend) the OECD (2001: 39): The intuition that "society matters" is strong and can be supported by much in the way of theoretical

16 Lars Osberg underpinning.' For economists and policy makers, of all stripes and at all levels, the task that remains is achieving a more precise estimation of the size of the impacts of social cohesion on economic performance. The general conclusions of this volume are four: (1) Social cohesion has significant economic implications. (2) Social cohesion has the potential for both positive and negative consequences. (3) The occasionally conflicting evidence on overall relationships may arise because highly aggregated measures (such as 'trust' or 'associational activity' in general) merge together the positive and negative aspects of social cohesion, which underscores the importance for social and economic policy of research which distinguishes the dimensions of social cohesion, and unpacks their different influences. (4) An understanding of all of the consequences and implications of social cohesion is going to be increasingly important for Canada. Notes 1 Interestingly, Kuznets suggested (in 1966) that in three to five decades the communist countries, having constructed their basic capital framework, would evolve into 'a typical welfare state with a broad democratic framework' (1966: 508). 2 Thanks to my colleague Mel Cross for this citation, and others similar. 3 In July 2001, the Econolit database had 408 hits on the term 'social capital,' only 46 of which date from 1995 or before. 'Social cohesion' had 76 hits, 25 from 1995 or before. 4 Coleman argued that social capital is highly productive and identified three forms: 'obligations and expectations ... information flow capability of the social structure and norms accompanied by sanctions. A property shared by most forms of social capital that differentiates it from other forms of capital is its public good aspect: the actor or actors who generate social capital ordinarily capture only a small part of its benefits, a fact that leads to under-investment in social capital' (1988: SI 19). 5 The awareness in international agencies of the importance of social cohesion may also be heightened by the current necessity for international meetings, such as the Organization of American States (OAS), G7, and the World Trade Organization (WTO) to occur in clouds of tear gas, behind phalanxes of police. The protestors opposed to the 'Washington consensus' that these meetings embody of domestic labour market 'flexibility' and increased 'harmonization' of trade practices are walking evidence of a lack of social cohesion to support this policy agenda. 6 To be clear about my own view, I think that human capital analysis combines an investment theory of individual motivation and a marginal productivity theory of

Introduction

17

market returns. Implicit in the Becker definition is the hypothesis that individual labour market returns can only increase when individual productivity increases. This implicit assumption is important in, for example, distinguishing between a human capital and credential model of the impact of education on earnings. If years of education is primarily a credential enabling individuals to move up a job queue, investment in more schooling may be privately profitable, but there is no social return. A human capital model argues that education is both socially and privately productive. My own preference would be to define human capital investment as "costly activities that influence future monetary and psychic income by increasing the productive resources in people.' This definition would address the social and private return issue which the human capital literature often now evades, but it must be recognized that my definition is narrower than most. 7 Benjamin, Gunderson, and Riddell (1998: 303), for example, have stated: 'Labour supply also has a quality dimension encompassing human capital elements such as education, training, labour market information, mobility and health' - which leaves open what other elements might be counted as part of human capital. 8 Costa and Kahn (2001) argued that the decline in U.S. community social capital from 1952 to 1998 was largely the result of rising income inequality.

References Becker, G. 1964. Human Capital. New York: Columbia University Press. Benjamin, D., M. Gunderson, and W.C. Riddell. 1998. Labour Market Economics: Theory, Evidence and Policy in Canada, 4th ed. Toronto: McGraw-Hill Ryerson. Coleman, J.S. 1988. 'Social Capital in the Creation of Human Capital.' American Journal of Sociology 94 (Suppl.): S95-S121. Costa, D., and M.E. Kahn. 2001. 'Understanding the Decline in Social Capital, 19521998.' Working Paper 8295. Cambridge, Mass.: National Bureau of Economic Research. De Tocqueville, Alexis. 1961. Democracy in America, vol. 2, chapter 8. New York: Schoken Books. Knack, Stephen, and Philip Keefer. 1997. 'Does Social Capital have an Economic Payoff? A Cross-Country Investigation.' Quarterly Journal of Economics Nov.: 1251-88. Kuznets, S. 1966. Modern Economic Growth: Rate, Structure and Spread. New Haven, Conn., and London: Yale University Press. OECD. 2001. The Well-being of Nations: The Role of Human and Social Capital. Paris: OECD. Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press.

18 Lars Osberg - 2000. Bowling Alone: The Collapse and Revival of the American Community. New York: Simon and Schuster. Schumpeter, J.A. 1950. Capitalism, Socialism and Democracy, 3rd ed. New York: Harper Torch Books, the University Library, Harper and Rowe Publishers. Smith, Adam. 1986. The Theory of Moral Sentiments, 6th ed. London: A. Millar. Reprinted in The Essential Adam Smith ed. by Robert L. Heilbroner. New York: WW Norton. Stanley, Dick. 1997. 'The Economic Consequences of Social Cohesion.' Hull, Que.: Department of Canadian Heritage, SRFA-302.

Do Borders Matter for Social Capital? Economic Growth and Civic Culture in U.S. States and Canadian Provinces JOHN F. HELLIWELL1

This chapter extends to North America the analysis of Helliwell and Putnam in their 1995 study linking regional growth and social capital in Italy. In the United States and Canada, there are also strikingly large and long-lasting regional differences in trust and other measures of social capital, and these have, as in Italy (Putnam 1993), roots far in the past. This is more surprising for the United States and Canada, where migration has been much larger and more varied than in Italy. U.S. census and survey evidence suggests that immigrants and internal migrants tend to bring trust or mistrust with them and to pass it on to their descendants (Fischer 1989; Rice and Feldman 1997). What are the implications, if any, for economic growth? There is theoretical and empirical support for the idea that societies whose members have high levels of mutual trust can operate markets and other economic and social institutions with fewer disputes and lower transactions costs (e.g., Platteau 1994a; 1994b). Should this be expected to show up as higher levels or rates of growth of gross domestic product (GDP) per capita? For Italy, there is some evidence that regions with higher levels of social capital have more effective regional governments, and grow faster, subject to the forces of convergence permitting the lagging regions to learn from better practices elsewhere in the country. Among the U.S. states (Barro and Sala-i-Martin 1991) and among the Canadian provinces (Helliwell 1996b), however, the extent of convergence over the past decades has been so strong that it may be difficult to identify a growth-inducing role for social capital, especially as measures of social capital tend to be higher where average incomes are higher. This chapter lays out the evidence and documents the extent and nature of the interprovincial, interstate, and international differences in social capital, using a number of comparable measures from the World Values Survey (WVS) of 1990-1. (Inglehart et al. 1995). Attempts are also made to see to

20 John F. Helliwell what extent Canada has suffered recent declines in social capital of the sort documented by Putnam (1995) for the United States. Are national borders as important for measures of social capital as they are for trade flows (McCallum 1995; Helliwell 1996c; 1998)? Comparable measures of social capital for U.S. states and Canadian provinces should help to answer this question. The next section of the chapter provides a comparative analysis of differences among provinces, among states, and between provinces and states, in key aspects of social capital. The following section assesses the linkages between these and some measures of economic performance. The essay is exploratory in nature, partly because of the shortage of comparable data, and partly because there is not enough variation over time in the measures of social capital to unravel the undoubtedly complex two-way linkages between social capital and economic performance. It is also early days in the development of cleanly distinct hypotheses about how political and social institutions and norms should be expected to influence individual and aggregate economic performance, and vice versa. Social Capital in U.S. States and Canadian Provinces What are the most relevant ways of measuring those aspects of the political and social system that might influence the operation of a modern industrial democracy? In their pioneering comparative empirical study of post-Second World War democracies, Almond and Verba (1963) 'concluded that interpersonal trust is a prerequisite to the formation of secondary associations, which in turn is essential to effective political participation in any large democracy' (Inglehart 1990). Over the forty years since their empirical work in the 1950s, researchers in many countries have probed the extent of interpersonal trust and examined the degree to which individuals participate in, and contribute their efforts to, voluntary associations. There have also been attempts to establish the extent to which interregional measures in these variables, described sometimes as measures of social capital or of civic culture, are linked with the extent to which citizens in the regions of a country are satisfied with the efficiency of their regional governments (Putnam 1993), and in turn whether there are subsequent pay-offs in terms of conditionally higher growth in those regions with higher levels of social capital (Helliwell and Putnam 1995). For the regions of Italy, where detailed research was carried out for more than twenty years, the results were fairly supportive of the notion that there was a causal linkage running from high levels of trust and engagement to higher levels of regional government performance and, in turn, to higher levels of

Do Borders Matter for Social Capital?

21

economic performance and higher rates of upward convergence towards best practice levels of efficiency. This invites consideration of whether there are similarly large and longstanding regional differences of social trust and civic engagement within other countries. The issues of regional differences and national trends could then be placed in comparative context, with an eye to assessing their relative size and significance. At a second stage, if these variations among regions and over time should be found significant, their importance in the broader analysis of political and economic life would need to be established. Finally, the use of individual responses should help to reveal the possible importance of inequality or other social cleavages on average levels of trust and participation, a matter not dealt with in earlier studies based on national averages of responses. This section will make use of data from the World Values Survey (WVS), using individual response data from the United States and Canada from two rounds of surveys, one undertaken in 1981 and the other in 1990. Although the WVS has the great advantage of asking the same questions in both countries, at about the same dates, there are three main difficulties: (1) The restriction to only two dates makes changes over time difficult to identify. (2) There are anomalous differences between the WVS and other measures of trust, especially in 1990. (3) Although the WVS records data for all Canadian provinces, the U.S. states are classified only in groups (regions), and these groups do not correspond fully to the lines of cleavage that appear among states in other data. 'Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?' Many thousands of interviewees have been asked this question over the past forty years, in scores of countries and languages. To provide something between the two alternatives, for those who might agree with both or neither, respondents could answer 'Don't know.' A value of 1.0 is coded for the trust variable for those who think that most people can be trusted and 0 for those who think instead that 'You can't be too careful.' When Almond and Verba (1963) took their 'snapshot' of civic culture in five countries in the late 1950s, they asked this question to citizens of five countries. They were struck by how much higher social trust appeared to be in Great Britain and the United States than in Germany and France. Almond and Verba saw these differences as crucial, since they considered interpersonal trust to be a prerequisite to the formation of secondary associations, with the latter in turn being essential to effective political participation in any large democracy. As more recent evidence from

22 John F. Helliwell the World Values Survey has shown (Inglehart 1990), this gap has been narrowed, with convergence being achieved partly through increases in trust in Germany and Italy and partly by reductions in trust in the United States and Great Britain. This narrowing of the gap in social trust has been matched by a narrowing of the postwar gaps in per capita incomes. Putnam (1995) has documented the U.S. decline in several measures of social capital, including social trust and participation in a variety of voluntary organizations. Putnam (1993) has since attempted to whittle down the number of plausible contributing factors, leaving television as the principal culprit. Subsequent research has emphasized the differences among types of organization, with some growing and others declining, and with rising educational levels doing much to offset other factors leading to a decline in participation. Despite the downward trend in the national average levels of trust, Rice and Feldman (1997) have shown a remarkable continuity from generation to generation in trustfulness, with American descendants of European migrants differing among themselves in their trust levels in ways that mimic contemporary differences among the European countries from which their parents or grandparents departed, even after two or more generations in the melting pot. The World Values Survey contains the same trust and membership questions in both Canada and the United States. The equations in Tables 1 and 2 analyse the results for trust, while Tables 3 through 5 do the same for memberships. In Tables 1 and 2 the equations are estimated, first, for the 1,576 U.S. observations, and then for 1,634 Canadian observations, and finally for the pooled sample containing data from both countries. Table 1 looks for differences by province or region, while Table 2 asks whether there are differences in trust corresponding to the ethnic categories that people use to describe themselves. There are two versions for each trust equation, one with and one without a variable representing the individual's participation in a variety of organizations. Both versions include a variable measuring the age at which the respondent finished full-time education, with an increase of 1.0 representing an additional year of schooling.2 The effects of post-graduate education are conflated, since the highest value of the variable, 10.0, is used for all those who finished full-time education at the age of 21 or older. The educational effects are strongly significant in both versions of the equation, but are higher in the equations without the number of memberships included. This fortunately does not pose any ambiguity of interpretation, since the equations in Tables 3 and 4 show that memberships are themselves influenced by education, in an amount that makes the implied effects of education

Do Borders Matter for Social Capital?

23

Table 1 Trust in U.S. and Canadian regions, 1990 Equation

Country Observations Estimation method Dependent variable Constant

Coefficients Age leaving school Memberships South WNorth-Central

(1)

(2)

(3)

(4)

(5)

(6)

U.S. 1576 OLS Trust 0.281 (6.0)

U.S. 1576 OLS Trust 0.274 (5.8)

Canada 1634

Canada 1634

Both 3210 OLS Trust 0.232 (7.1)

Both 3210 OLS Trust 0.239 (7.3)

0.024 (4.0) 0.031 (5.0) -0.073 (2.4) 0.105 (2.5)

0.032 (5.7)

0.031 (8.1) 0.026 (6.1) -0.071 (2.4) 0.110 (2.7) -0.128 (4.7) 0.180 (6.0) 0.048 (2.1)

0.037 (9.9)

-0.077 (2.6) 0.132 (3.2) -0.135 (5.0) 0.176 (5.8) 0.044 (1.9)

0.073 0.481

0.063 0.483

OLS

OLS

Trust 0.248 (6.2)

Trust 0.258 (6.5)

0.036 (7.3) 0.022 (3.6)

0.040 (8.2)

-0.080 (2.6) 0.131 (3.2)

QU+NB AL+BC

-0.175 (6.3) 0.131 (4.3)

-0.178 (6.4) 0.132 (4.3)

0.098 0.474

0.092 0.476

Rest of Canada

Rz S.E.E.

0.048 0.487

0.033 0.491

the same whether one looks at the trust equation as a reduced form (thus excluding the membership variable) or treats the two equations as a system, substituting the estimated equation for memberships into the trust equation to calculate the direct plus indirect effects of education on the level of trust. A sample calculation may help to illustrate this result, as well as to show the important effects of education on the level of trust. The first equation in Table 1 shows that every additional year of education increases by 0.024 the probability that a U.S. respondent thinks that most people can be trusted, while adding a new membership category raises the probability by 0.031. From Equation (5) in Table 3, however, we can find that another year in fulltime education increases memberships by 0.264, so that the total effect, including the direct effects on trust and the indirect effects acting through membership activity, are 0.032 (= 0.024 + 0.031 * 0.264). In the second equa-

24 John F. Helliwell tion, by contrast, there is no direct membership effect, and the reduced-form effect of education is directly estimated as 0.032, the same value that results from solving the two equations for trust and memberships. Postulating that memberships influence trust, rather than the other way around, is consistent with the results of Brehm and Rahn (1997), who found the apparent influence much stronger from memberships to trust rather than vice versa. More generally, however, it is reasonable to suggest, following Putnam (1995), Brehm and Rahn (1997), and others, that trust and memberships are positively reinforcing, with membership contacts helping to foster trust, and higher trust levels encouraging more membership activity. Thus, it is important to estimate the equations with neither appearing in the other's equation, to provide a reduced-form estimate of the full influence of education, region, ethnic affiliation, or other predetermined variables influencing both trust and memberships. The total effects of education are large in both countries, but larger in Canada than in the United States. The total effect in the United States is 0.032, compared with 0.040 in Canada.3 A four-year undergraduate degree would raise an expected trust level of 0.40 to 0.53 in the United States or to 0.56 in Canada. The lower part of Table 1 shows the most important regional variations in trust in the two countries. In each country there is one region with sharply lower trust and another with sharply higher trust. In the United States, trust is about 0.08 lower in the South and 0.13 higher in the West North-Central region, with the remaining regions treated as the base for comparison. In Canada, trust is about 0.18 lower in Quebec and New Brunswick and 0.13 higher in Alberta and British Columbia, once again in comparison with the rest of the country (ROC). These calculations are based on the reduced-form equations without including the effects of membership, which also has some regional differences, although these are much less marked than those for trust. Equations (5) and (6) in Table 1 combine the U.S. and Canadian data to permit an evaluation of whether there are significant national differences in the level of trust. Once again, there are equations both with and without the intervening effects of membership, and for simplicity the discussion below will concentrate on the reduced-form equations without membership included. The explanatory power of the Canadian equations is higher than that of the U.S. equations, whereas that of the combined equation naturally falls about midway between the two. The systematic regional effects in each country are found again in the two-country sample, and a new variable covering the rest of Canada is added to see if there are significant trust differences between the two countries.

Do Borders Matter for Social Capital?

25

The coefficient suggests that trust levels for individuals living outside the areas of specially high or low trust are 0.044 higher in Canada than in the United States, assuming the same number of years of education in the two cases. Predicted trust levels generally rise as one crosses the border from south to north, but there are important exceptions, and the regions used for subdivision have no cast-iron claim to capture accurately the boundaries of trustful or distrustful communities. For example, since Minnesota is in the high-trust West North-Central region, and Manitoba is a part of baseline Canada, predicted social trust drops as one crosses the border to the north, as it does if one crosses from New England into Quebec. However, it rises slightly crossing from New York into Ontario, and much more moving from Washington State into British Columbia.4 Why does trust vary so much from region to region, and generally rise from south to north-central in the United States and from east to west in Canada? Do open spaces and cold winters combine to make people rely on each other, as many of the pioneer histories would suggest? Does the pacifying influence of the early imposition of the rule of law, rather than of force of arms, help to explain higher trust north of the forty-ninth parallel? What is likely is that the patterns of trust in 1990 have their origins many years in the past, if Putnam's Italian research can be transposed to North America. The big difference, of course, is that the western parts of North America underwent their major population growth only in the twentieth century. What was special about some migrations, or about some features of new communities, that made them more likely to have higher levels of social trust? Specific answers to questions of this sort will require more information than is in the current sample data, since there is too little information there about the type and origins of the communities in which the individual respondents live. What is very likely, however, is that much of the difference among individuals in their degree of trust will be found to depend on the extent of trust in the communities from which they migrated, even if that migration should be long past. Perhaps part of the reason for the high level of trust in Minnesota (0.63 compared with the national average of 0.42) may be the cold climate and the social life of the hockey rinks, but more important, according to recent research by Rice and Feldman (1997) is that one-third of the population of Minnesota is of Norwegian extraction, and Norway has, according to comparable WVS data, one of the highest trust levels in the world. If trust levels are strongly persistent, as their finding that grandparent effects are almost as important as parent effects would suggest, then the high trust levels in Minnesota were brought with them by the Norwegian immigrants. What

26 John F. Helliwell remains to be discovered is what features of subsequent community life and development were helpful, and which were not, in maintaining or improving the imported levels of trust. Perhaps a similar story can be told for Quebec, as the trust levels in France are well below those of other Western European countries, and well below those in Quebec. However, it must be remembered that the pur laine Quebecers left France long before the French Revolution and have had a longer and more separate existence than any of the comparable ethnic groups in either Canada or the United States. One of the difficulties of assessing the importance of migration as a determinant of trust and other forms of social capital is that although state and provincial boundaries were, in part, drawn to reflect previous patterns of migration, subsequent migration patterns have been so complicated and varied that state and provincial averages, let alone the multistate regions covered by the WVS, are likely to contain too much heterogeneity of origin and contemporary community structure. The burden of the story of Rice and Feldman is that the melting pot does not melt very fast, even in the migratory United States, but it is almost certain that a finer grid than states or nations is needed to assess the way the process works. There is also a problem with too much aggregation in the description of the source of migration. Fischer (1989) has argued convincingly that four successive waves of English migration came with strikingly different types and amounts of social capital and maintained their distinct cultures over the subsequent centuries. The contrasts he draws between the Puritan migrants from East Anglia to Massachusetts, the southern English cavalier migrants to Virginia, the Quaker migrants to Delaware, and the borderlands migrants to the backcountry are extreme, and their shadows far-reaching. For example, each of the initial groups played separate but important roles in opening up the west, and in determining the type of social capital to be found in the communities they established there,5 twice transplanted from their source communities in Britain. The Quakers brought town meetings and a tradition of ordered liberty to Massachusetts. The Virginia settlers brought hierarchies and 'strong oligarchies, Anglican churches, a highly developed sense of honour and an idea of hegemonic liberty' (Fischer 1989: 786). The Friends brought to Delaware a work ethic, austerity, and 'a pluralistic system of reciprocal liberty.' The migration from the border country to the backcountry was later and much larger than its predecessors, and the migrants themselves were very different from those who preceded them. These forebears of Patrick Henry were leaving hard conditions and hoping for better. They were used to the

Do Borders Matter for Social Capital?

27

endemic violence of the border regions, and they brought with them suspicion of formal education and authority, distinctive modes of dress and speech, rituals of love and violence, militant Christianity, clan loyalties, and violence which came to mark the settlements they founded in 'the backcountry region that included southwestern Pennsylvania, the western part of Maryland and Virginia, North and South Carolina, Georgia, Kentucky and Tennessee' (Fischer 1989: 634). The four cultures described by Fischer may all have come from England, but they embodied vastly different amounts and types of social capital, were generally antipathetic to each other, and developed communities where they settled, and where they subsequently moved, that had little in common with each other, although the new communities often bore striking resemblances to their own predecessors. What can we conclude from this? Most importantly, it is clear that using nations as definitions of source populations, and states or provinces as the units for analysis on this side of the water, is bound to cover up as much as it reveals. It is perhaps surprising that regional differences have even managed to show up at all. What can be done to further cast light on the possible differences among individuals in their trust and involvement? Table 2 attempts an alternative classification that may help to disentangle these differences. What was done here was to divide respondents according to how they classified themselves. The categories used for the United States were Hispanic American, Black American, Asian American, and White American. The base alternative, chosen by 28 per cent of the respondents, was 'American first, and then member of some ethnic group.' For Canada, the three hyphenated categories were French Canadian, English Canadian and Ethnic Canadian. The base alternatives were 'Canadian first, and then member of some ethnic group' (13 per cent of respondents) or 'Canadian first and only' (38 per cent of respondents). In both countries, those who defined themselves with an ethnic qualifier were less likely to think that people could be trusted. The equations all included separate allowance for the effects of education and were estimated with and without the inclusion of the memberships variable. The ethnicity effects are larger in the equations without memberships included, reflecting the fact that those who qualify their citizenship by ethnicity also have lower than average membership activity. The biggest departure from baseline trust is among those who describe themselves as Black Americans (0.32 below baseline), followed by French Canadians (0.22 below baseline) and English

28

John F. Helliwell

Table 2 Trust in the U.S. and Canada by affiliation, 1990 Equation

Country Observations Estimation method Dependent variable Constant

Coefficients Age leaving school Memberships Hispanic American Black American Asian American White American

(D

(2)

(3)

(4)

(5)

(6)

U.S. 1576 OLS Trust 0.349 (6.8)

U.S. 1576 OLS Trust 0.357

Canada 1634 OLS Trust 0.307 (7.6)

Canada 1634 OLS Trust 0.317 (7.8)

Both 3210 OLS Trust 0.281 (7.5)

Both 3210 OLS Trust 0.300 (9.3)

0.037 (7.5) 0.023 (3.7)

0.041 (8.4)

0.030 (7.9)

0.036 (9.3)

-0.107 (1.7) -0.293 (6.1) -0.137 (1.0) -0.033 (1.2) -0.142 (3.3) -0.114 (3.8) -0.072 (1.2) 0.073 (2.7)

-0.135 (2.2) -0.317 (6.6) -0.165 (1.2) -0.044 (1.6) -0.162 (3-8) -0.131 (4.3) -0.079 (1.4) 0.059 (2.1)

0.074 0.481

0.064 0.483

0.021 (3.6) 0.030 (4.9) -0.114 (1.8) -0.299 (6.2) -0.136 (1.0) -0.037 (1.3)

(6.9) 0.028 (5.1)

-0.144 (2.3) -0.324 (6.7) -0.167 (1.2) -0.048 (1.8)

French Canadian English Canadian Ethnic Canadian

-0.210 (5.3) -0.187 (7.1) -0.146 (2.6)

-0.216 (5.5) -0.189 (7.1) -0.139 (2.5)

0.087 0.477

0.080 0.479

Other Canadians

R2 S.E.E.

0.062 0.484

0.048 0.488

Canadians (0.19 below), but the effects are negative, and generally significantly so, for all seven categories, comprising four in the United States and three in Canada. Equations (5) and (6) combine the data for the United States and Canada, permitting the trust levels in the two countries to be compared directly. Comparing unhyphenated Canadians and Americans of the same educational level, trust levels in 1990 were about 0.06 higher in Canada than in the United States. The difference is slightly larger if one compares people with the same

Do Borders Matter for Social Capital?

29

membership activity also (using Equation 5), since memberships are generally higher in the United States. The Table 2 results by ethnic classification are quite striking. They suggest that trust levels are lower where perceived cultural cleavages are stronger, if it is reasonable to assume that those who classify themselves with an ethnic qualifier do so because of perceived cleavages. This effect is as apparent among those who describe themselves as English Canadian as it is among those who describe themselves as French Canadian, suggesting that the perception of cleavage is as important as whether one is in the majority or the minority. The baseline Americans or Canadians who think of themselves first as Americans or Canadians, and then later, or not at all, as members of some ethnic group, are more trusting than their fellow citizens. Perhaps the bridging institutions and experiences that helped them to maintain a higher level of trust also made them more likely to emphasize the common elements rather than their differences in their descriptions of themselves.6 In the previous analysis all membership types have been treated equally. Is it true that membership activity is all that counts, or are some types of voluntary organizations more likely than others to build bridges of the sort that encourage social trust and other forms of social capital? Table 3 and 4 show regional and ethnic differences in memberships of various types for the United States and Canada separately, and Table 5 compares the differences between the United States and Canada in memberships of different types. The sixteen types of voluntary organization are aggregated into four main groupings for analysis: (1) sports and cultural, (2) political, (3) religious, and (4) trade union.7 Looking first at Table 3, showing U.S. memberships by type, education, and ethnic grouping, several main features are apparent in the results. Educational effects are significantly positive for all types of membership except trade unions, where there is no effect. The positive educational effects are especially high for sports, cultural, and political organizations. Turning to the regional effects, trade union and sports and cultural memberships are significantly lower, and religious memberships slightly higher in the South. In the West North-Central region, memberships are much higher in all types of organization except trade unions, where the national average prevails. All of the four ethnic groups are significantly less likely to participate in sports and cultural and political organizations, while not differing significantly from national average memberships in religious and trade union groups. In Canada, the educational effects are significantly positive for all but religious organizations. The only significant regional differences in memberships are for religious groups, where memberships are significantly below

30 John F. Helliwell Table 3 U.S. Memberships by type, 1990 Equation

0) 1576 observations Average no. of memberships Estimation method Dependent variable Constant

Coefficients Age leaving school South WNorth-Central Hispanic Black Asian White

R2 S.E.E.

(2)

(3)

(4) 0.09

(5)

2.01

0.61

0.49

OLS Sports+cult 0.903 (17.1)

OLS Political 0.777 (16.1)

OLS Religious 0.484 (19.7)

OLS OLS Trade Union Total 2.25 0.086 (6.0) (23.0)

0.119 (9.5) -0.141 (2.1) 0.306 (3.4) -0.342 (2.5) -0.304 (2.8) -0.451 (1.4) -0.130 (2.1)

0.120 (10.5) -0.051 (0.8) 0.300 (3.6) -0.552 (4.4) -0.375 (3.8) -0.582 (2.0) -0.271 (4.9)

0.023 (4.0) 0.050 (1.6) 0.197 (4.7) -0.092 (1.4) -0.019 (0.4) 0.072 (0.5) -0.043 (1.5)

0.002 (0.7) -0.060 (3.3) 0.016 (0.6) -0.008 (0.2) -0.012 (0.4) -0.083 (1.0) 0.037 (2.3)

0.264 (11.4) -0.202 (1.6) 0.818 (4.9) -0.993 (3.9) -0.711 (3.6) -1.043 (1.8) -0.407 (3.6)

0.081 1.07

0.106 0.975

0.023 0.494

0.012 0.288

0.121 1.97

0.81

national averages in both the special regions, more so in Quebec and New Brunswick than in Alberta and British Columbia. Among the ethnic groupings, there are no significant membership differences, except that those calling themselves ethnic Canadians are significantly more likely to be involved in political organizations. Overall, there is much less regional and ethnic variation of memberships in Canada than in the United States. Comparing Canada and the United States more directly, as is done in Table 5, shows that Canadians of similar educational levels are much less likely than are U.S. respondents to belong to religious organizations and more than one-third more likely to belong to trade unions.8 The pattern of educational effects differs significantly by type of organization. Educational effects for sports and cultural organizations and for political organizations are about equal, although both are about half again as large in the United States as in

Do Borders Matter for Social Capital? 31 Table 4 Canadian memberships by type, 1990 Equation

(1) 1634 observations Average no. of memberships Estimation method Dependent variable Constant

Coefficients Age leaving school French Canadian English Canadian Ethnic Canadian QU+NB AL+BC R2 S.E.E.

0.80

(2)

0.52

(3)

0.25

(4)

0.12

(5)

1.69

OLS

OLS

OLS

OLS

Sports+cult 0.811 (19.3)

Political 0.494 (13.3)

Religious 0.331 (19.1)

Trade union Total 0.121 1.758 (22.7) (9.1)

0.079 (7.4) 0.007 (0.1) -0.010 (0.2) 0.087 (0.7) 0.007 (0.1) 0.053 (0.8)

0.078 (8.3) -0.077 (0.9) -0.033 (0.6) 0.304 (2.9) 0.064 (1.0) 0.112 (1.9)

0.002 (0.4) -0.004 (0.1) -0.034 (1.3) -0.079 (1.6) -0.155 (5.6) -0.109 (4.0)

0.012 (3.6) -0.039 (1.3) 0.024 (1.2) -0.011 (0.3) 0.019 (0.9) -0.022 (1.0)

0.171 (8.7) -0.192 (1.1) -0.054 (0.5) 0.302 (1.4) -0.065 (0.5) 0.034 (0.3)

0.032 1.034

0.048 0.912

0.026 0.427

0.009 0.326

0.047 1.910

OLS

Canada.9 For religious groups, there is a much smaller, but still significant, positive effect of education in the United States, but no such effect in Canada. For trade unions, there is no significant educational effect in the United States, but a modest and significant positive effect in Canada. This latter difference is no doubt related to the fact that service industries, and especially the public service, where educational levels are above national averages, are relatively much more unionized in Canada than in the United States.10 Thus, we see that the higher average memberships in the United States are concentrated in religious organizations, where Canadians are half as likely as U.S. respondents to belong. Americans are also somewhat more likely to belong to political organizations, while memberships in sports and cultural organizations are the same in both countries. Trade unionism is the only form of membership that is significantly higher in Canada than in the United States. Given the quite different patterns of memberships in the two coun-

32 John F. Helliwell Table

5 U.S. and Canadian memberships compared, 1990 Equation

0) 3210 observations Average no. of memberships Estimation method Dependent variable Constant Age leaving school (U.S.) Age leaving school (Canada) Canada

R2 S.E.E.

(2)

(3)

(4)

(5)

0.56

0.37

OLS Sports+cult

OLS Political

OLS Religious

OLS OLS Trade union Total

0.785 (29.4) 0.128 (10.6) 0.081 (7.5) 0.033 (0.9)

0.584 (24.2) 0.131 (12.0) 0.081 (8.2) -0.052 (1.5)

0.486 (41.3) 0.023 (4.3) 0.002 (0.5) -0.237 (14.4)

0.092 (11.8) 0.003 (0.9) 0.012 (3.7) 0.033 (3-0)

1.95 (39.4) 0.286 (12.8) 0.176 (8.8) -0.224 (3.2)

0.049 1.05

0.063 0.952

0.069 0.465

0.006 0.309

0.074 1.95

p=.004

p=.069

p=.0002

0.81

Probability of schooling p=.003 effects being the same in Canada and United States

p=.0005

0.11

1.85

tries, it is of interest to ask whether some types of membership are more likely than others to contribute to social trust. The answer, in general, is no. If the basic equations with trust explained by education and memberships are estimated with the four different types of memberships separately and together, there are no differences by type of membership for the United States, and the equation fits better with the total than with any one or subgroup of membership types. For Canada, the membership total excluding trade union membership has a slightly higher explanatory power, since the contribution of the trade union component is insignificantly negative, but the difference is not significant. If it is true that trust is built most by involvement in bridging rather than separating institutions, then it is perhaps not surprising that trade union membership is less likely to increase general trust, since unions build solidarity among workers in part by emphasizing a cleavage between workers and

Do Borders Matter for Social Capital? 33 managers and between capital and labour. With that slight and partial exception, it seems generally appropriate to conclude, as other researchers have argued, that membership in almost any organization helps to develop knowledge and shared values that contribute to attitudes of trust. How do changes in social capital compare in the two countries? On this important question the data pose difficulties. The only really comparable data for the two countries, collected at similar times, come from the World Values Surveys of 1981 and 1990. The same questions were asked in both countries, and it is possible to factor out the possible confounding role of educational differences in the two samples. There are two main problems. One is that there are only two data points for each country. The second and more fundamental problem is that the 1990 WVS data show values of trust that are much higher than those from the General Social Survey (GSS) and National Election Survey taken at nearby dates, and higher than was found by the WVS in 1981. Thus, the WVS shows trust rising in the United States from 1981 to 1990, and this makes more difficult the comparison between Canada and the United States. If the data from both surveys in both countries are pooled together in one large sample, with each of the other trust levels compared with that in the United States in 1990, an equation with educational levels included shows U.S. trust in 1981 to be 0.10 lower than in 1990. Canadian trust in 1981 is equal to U.S. 1990 trust, while Canadian 1990 trust is 0.02 higher. Looking just at the sample average levels of trust, which is not unreasonable since the educational differences are not very large (educational levels are roughly six months higher in the United States than in Canada, in both years, and are about six months higher in 1990 than in 1981, in both countries), Canadian trust levels rose from 0.48 in 1981 to 0.53 in 1990, while those in the United States rose from 0.40 in 1981 to 0.52 in 1990. The WVS data thus show trust levels to be higher in Canada than in the United States, by a much bigger margin in 1981 than in 1990. Whether these data should be taken to imply a drop in trust in Canada, if other evidence points to a drop in the United States, is not clear because the answer would depend on how many of the four samples were thought to be anomalous. The membership data also show increases from 1981 to 1990 in both countries. It may be possible to locate and standardize for sampling differences that make the U.S. 1990 WVS data for trust and memberships different from other U.S. evidence. In the meantime, the conclusion has to be that we simply cannot yet tell from the WVS data whether social capital has declined in Canada over the 1980s, or how any Canadian changes relate to those in the United States.

34 John F. Helliwell Economic Growth and Social Capital If there is a positive linkage running from measures of social capital to regional economic growth in Canada and the United States, it will be difficult to uncover. In the United States, there has been a striking convergence in the levels of per capita incomes over the past century, and the same has been true for Canadian provinces, at least over the post-1926 period for which comparable data are available. Economic growth has been fastest in those states and provinces that had the lowest per capita incomes at the beginning of the period. But the results in the preceding sections show that these are also generally the regions with lower than average levels of trust and, to a lesser extent, of memberships. For example, the U.S. Southern states have had significantly lower trust levels, while having the fastest rates of growth over the past century. The two most western Canadian provinces have had higher trust and memberships and low rates of growth of per capita incomes. This does not mean that social capital is bad for growth, as the Italian study showed a positive partial effect of social capital on growth even though the poorer regions grew fastest and had the lowest levels of social capital. The faster growth of the poorer regions is explained by conditional convergence, wherein, once certain conditions were established, the poorer regions grow faster because as they adopt more efficient techniques already developed and in use in other regions. The hypothesis is that regions with higher levels of social capital may be able to coordinate and implement the required changes in institutions and performance more easily and at less cost because trust and informal cooperative networks can both facilitate the passage of ideas and make less costly the system of contracts and contacts required to support innovation and growth (Platteau 1994a, 1994b). This hypothesis was supported by the Italian results (Helliwell and Putnam 1995). Among the industrial countries, however (Helliwell 1996a), and among U.S. states and Canadian provinces, the convergence has been so rapid as to obscure any simple positive effect flowing from levels of trust and membership to subsequent economic growth. This result may change with better data and analysis and as the longer and more dramatic convergence process starts to slow down. In any event, it may well remain true, in North America as it is in Italy, that the regions with higher levels of social capital retain higher than average income levels even after many years of convergence. What do the data show so far? If we focus especially on the period since 1950, the coefficient of variation of per capita personal incomes among U.S. states fell from 0.22 in 1950 to 0.19 in 1960 and 0.14 in 1980. In Canada, the dispersion and convergence

Do Borders Matter for Social Capital?

35

patterns were very similar to those in the United States, at least until the U.S. dispersion started to increase again in the late 1970s. The coefficient of variation among provinces in personal disposable income fell from 0.21 in 1950 to 0.19 in 1960 and 0.15 in 1980. Since 1980, however, the experience in the two countries has been quite different, as is shown in Table 1. For U.S. states the coefficient of variation among state averages for personal income rose from 0.14 in 1979 to 0.17 in 1989, as previously noted by Crihfield et al. (1995). The similar statistic for personal disposable incomes rose from 0.13 in 1979 to 0.155 in 1989. In Canada, by contrast, the coefficient of variation of personal disposable incomes continued to fall, dropping from 0.136 in 1979 to 0.11 in 1989. Since 1990, the interstate income dispersion in the United States has started to fall once again, but is still substantially higher than in Canada. In both countries, the interregional variation of disposable incomes is smaller than that of pretax and transfer incomes, showing that the tax and transfer system does manage to make average post-tax incomes more equal than are pretax incomes. The gap between the two dispersion measures is quite similar in the two countries, despite the more extensive and more expensive Canadian system of social safety nets, including a much more equalizing system of payments from the federal government to provincial governments. Regional growth regressions for the United States, with time series for each of the nine census regions estimated as a system of equations, using the iterative Zellner SUR technique for estimated seemingly unrelated regressions, provide systematic evidence of convergence, as might be expected from the evidence shown in Table 1. However, adding measures of trust from the most recent previous GSS surveys does not add anything to the explanation of convergence. For Canada, there are provincial estimates of trust available only for 1990, and there is no correlation between these and the constant terms from provincial equations for the growth of per capita incomes. Thus, it still appears to be the case that there is no significant evidence that higher levels of social capital, at least as measured by trust, have a positive effect on the growth rates of per capita incomes. In addition to the growth or per capita incomes, there is also a plausible channel running from social capital to interregional or interprovincial migration. The hypothesis here is that measured levels of social capital provide indicators of a quality of life that may attract migrants. The positive correlations between levels of social capital and levels of per capita incomes may permit the measures of social capital to have an independent effect above and beyond the likely effect of per capita income differences as an incentive for interregional migration. If this is so, it would provide a direct linkage between

36 John F. Helliwell Table G Regional migration in the United States and Canada Equation

Estimation method Sample Time period Observations Dependent variable Relative personal income Trust

(D

(2)

(3)

(4)

(5)

SUR U.S. regions 1962-94 9x33 dln(pop)ft

POOL U.S. regions 1962-94 9x33 dln(pop)tt

POOL U.S. regions 1962-94 9x33 dln(pop)ft

SUR Cdn. provinces 1961-89 10x29 dln(pop)it

SUR Cdn. provinces 1961-89 10x29 dln(pop)K

0.0217 (6.0) 0.0163 (4.7)

0.0576 (6.1) 0.0103 (2.2)

0.0598 (6.4) 0.0056 (1.2) 0.00025 (4.9)

0.0467 (10.0)

0.0334 (8.0)

0.612

-0.0145 (17.2) 0.941

Time index Relative unemployment rate System R2

0.381

0.394

0.408

Notes: Absolute values of f-statistics are in parentheses below coefficients. The SUR estimations use the iterative Zellner procedure, and include constant terms for each region or province. The POOL results are estimated with separate intercepts (not reported) for each region, have separate autocorrelation coefficients for each region (ranging from 0.6 to 0.9), and allow for cross-sectionally correlated error terms. The time index takes the value 1.0 in 1961, 2.0 in 1962, and so on. The provincial constant terms from equations (4) and (5) are not correlated with the 1990 WVS measures of trust in the provinces.

social capital and regional growth because the regions with higher levels of social capital would attract population and economic activity, even without any induced change in average per capita incomes. Table 6 reports the results of systems of migration equations for the nine U.S. census regions and for the ten Canadian provinces. For both the United States and Canada, the dependent variable is the annual change in the logarithm of the population in a region or province. In both the United States and Canada, there is significant evidence that migration moves population from the poorer to the richer regions, with the magnitude of the effect being similar in the two countries. The Canadian results show that the difference in unemployment rates has an even more significant effect, suggesting the potential usefulness of extending the equation for U.S. migration rates to include the effects of

Do Borders Matter for Social Capital? 37 regional differences in unemployment rates. It is interesting that the migration rates within Canada appear to be as much or more dependent on economic influences than is migration in the United States, given other results suggesting that labour markets are more flexible in the United States than in other industrial countries, and given the much higher rates of interregional fiscal transfers in Canada than in the United States. In addition, the Canadian social safety nets are twice as expensive and have much broader and deeper coverage than those in the United States, leading one to expect that population might move more quickly in response to income and employment prospects in the United States than in Canada. That this does not appear to be the case may be surprising, but it is consistent with the earlier result that convergence of per capita incomes over the past thirty years has been faster in Canada than in the United States. The implication would seem to be that the higher degree of equity-based redistribution of incomes from the richer to the poorer regions has not substantially slowed down the migration and other economic adjustments that must eventually be relied upon to equilibrate economic opportunities over the longer run. Turning to the evidence on the impact of social capital, the U.S. equations suggest that migration does favour the regions with higher levels of trust, even after allowing for the impact of higher average incomes. Adding a time trend lessens the apparent effect, so the results should be regarded as provisional. No similar direct test is possible for the Canadian migration equations because there are no time series for the provincial measures of social capital. However, it is possible to see if the provincial constant terms from the migration equations are correlated with the provincial measures of trust for 1990. They are not. Thus, there is no apparent tendency in Canada, at least with the limited evidence available, for population to flow to provinces with higher levels of social capital, after allowing for the strong effects of differences in incomes and unemployment rates. Conclusions The chapter first assessed regional and ethnic group differences in social trust and memberships in both Canada and the United States, using the 1990 World Values Survey survey data in both cases. The effects of education on trust and membership are positive in both countries and larger in Canada than in the United States. In both countries, regions were found with significantly lower and higher than average levels of trust, with smaller echoes in membership activity. The ethnic categories people chose to describe themselves were as important as regional differences in explaining differences in trust. Re-

38 John F. Helliwell spondents who qualified their citizenship by any of seven adjectives (black, white, Hispanic, and Asian in the United States; French, English, and ethnic in Canada) had lower levels of trust than those who considered themselves Canadians or Americans either first or only. These categories also explained differences in memberships, but much more so in the United States than in Canada, where membership activities of the different types were more evenly spread across the country. It was argued that the regional differences in trust would have been easier to identify with smaller geographic units, given the apparent importance of past migration patterns which have varied greatly among and within states and provinces. Comparisons of membership activity between the United States and Canada showed the two countries to have very similar membership in sports, cultural, and political organizations, but very different patterns in religious and trade union activities, with the United States respondents being twice as likely to take part in a religious or church group and significantly less likely to be involved with a trade union. The WVS data were unable to show whether Canada has shared in the decline in trust and memberships over the past twenty years, since the WVS data actually suggest an increase in trust and memberships in both countries from 1981 to 1990, with the levels of trust being higher in Canada in both surveys, but the increase in trust being greater for the United States. More comparable data are required to explain the sources of the apparent contradiction between the diverging trends of the WVS and GSS data. Since there are many more observations by the GSS surveys, and these data match other surveys taken as part of the National Election Study (NES), it is likely that the resolution will be in favour of the GSS and NES data. If so, then the primary use of the WVS data will be in making international comparisons at a given time. Even so, the comparisons of the determinants of 1990 trust and membership in Canada and the United States must be treated with some caution, given the discrepancies that have been found between the WVS and other survey data. The chapter then turned to the linkages between measures of trust and economic growth. Comparable data for both personal incomes per capita and personal disposable incomes per capita show that the dispersion of incomes has been dropping in both countries, but faster in Canada than in the United States. In addition, the 1980s increase in regional income disparity in the United States had no parallel in Canada. In neither country was there evidence that per capita economic growth was faster in regions marked by high levels of trust. In this sense, the results of this chapter support those found

Do Borders Matter for Social Capital? 39 earlier for the member countries of the Organization for Economic Cooperation and Development rather than the results for the Italian regions. This discrepancy may be the result of the higher quality of social capital measures for the Italian regions, but even there the evidence was indicative rather than compelling. The chapter finally turned to consider whether the extensive margin of economic growth, as represented by migration from one province or region to another, was influenced by differences in social capital. In both countries there was significant evidence that migration is substantially triggered by differences in income levels. In the United States there was also evidence that, after adjusting for the effects of income differences, migrants tend to favour states with higher perceived levels of trust. Given the evidence that migrants tend to bring social capital with them, the induced migration is likely to play a role in changing the regional differences in social capital. The exact form of the relation cannot be specified without more knowledge about the characteristics of the migrants, and it is likely to depend also on the effects of the migration on the social and other institutional structures of the sending and receiving regions. It is simply too early in the research to conclude whether the net effect of the migration will be to cause a convergence of social capital of a sort that has characterized the distribution of income levels in both countries. Another interesting feature of the migration results was the finding that the responsiveness of migration appears to be, if anything, stronger in Canada than in the United States, despite the much more extensive systems of fiscal equalization and social safety nets in Canada. These programs, which are often thought to mirror a higher degree of social cohesiveness in Canada, are also thought by some to blunt the economic incentives to reallocate population and industry so as to equalize income opportunities across the country. Thus, it is encouraging to find, as these tentative results suggest, that the more extensive Canadian transfer programs have been accompanied by faster regional convergence of per capita incomes and by migration patterns that are highly responsive to differences in prospects for incomes and employment. Notes 1 Department of Economics, University of British Columbia, Vancouver B.C. V6T 1Z1. This paper was previously presented at the 'Policy Research: Creating Linkages' conference, Ottawa, 1-2 Oct. 1998. It is a revised version of a paper

40 John F. Helliwell

2

3

4

5

6

7

8

9

presented initially at the Annual Meetings of the Canadian Economics Association, Brock University, 31 May to 2 June 1996, and subsequently as NBER Working Paper, Dec. 1996. e-mail: [email protected]. This is the way the question is asked in the WVS survey, and the results are coded as 1 if the individual left school at the age of 12 or younger, 2 if 13, up to 10 if he or she left at age 22 or later. For the analysis in this paper, 6 is added to each observation to make the data approximately comparable with the U.S. GSS survey, which records the number of completed years of full-time education. As stop-andgo education becomes more common, this approximation will become less accurate, and the GSS form of the question will be likely to provide a more accurate measure of the level of educational attainment. A larger fraction of the U.S. total effect flows through the indirect membership channel than is the case for Canada, since the effect of education on memberships is lower in Canada, as is the estimated effect of memberships on trust. To be more specific, predicted trust falls by 0.09 from Minnesota to Manitoba (0.132 - 0.044), and by 0.13 from New England to Quebec. It rises by 0.044 from New York to Ontario, and by 0.22 from Washington to British Columbia (0.044 + 0.176). The back-country migrants later extended westward into Oklahoma and Texas, and many moved westward again to California at the time of the dustbowls of the 1930s. Roxanne Dunbar-Ortiz (1995: 572-3) reports that she knew what the verdict of the Rodney King trial would be as soon as she heard it was transferred to Simi Valley, which she knew from her own family history to be an 'Okie' town. These results are consistent with those of Kalin and Berry (1996), whose 1991 survey results show that those whose ethnic self-identity is as an unhyphenated Canadian, as distinct from those whose primary affiliation is to an ethnic group, are less ethnocentric and more accepting of multiculturalism. The sports and cultural grouping is the aggregate of six WVS groupings: social welfare, education, youth work, sports or recreation, health, and other. The political category is the aggregate of eight WVS categories: political parties, community action groups, world development groups, environmental groups, professional associations, women's groups, the peace movement, and animal rights. Churches and religious organizations are a separate WVS category, as are trade unions. The aggregate difference in unionization, while of relatively recent origin, is much larger than in the WVS sample. Riddell (1993: 110) reports that union membership as a percentage of non-agricultural paid workers, which in 1965 was at 30 per cent in both countries, had by 1990 fallen in the United States to 16 per cent and risen in Canada to 33 per cent. The importance of education as a factor increasing U.S. participation in political activities of many types is extensively documented in the survey results of Verba, Schlozman, and Brady (1995).

Do Borders Matter for Social Capital? 41 10 Using data from surveys of workers by the AFL and the CFL, Riddell (1993: 124) shows the effect of education on the demand for unionization to be higher in Canada than in the United States, but significantly so only for university graduates.

References Almond, Gabriel A., and Sidney Verba. 1963. The Civic Culture: Political Attitudes and Democracy in Five Nations. Princeton, NJ: Princeton University Press. Barro, R., and X. Sala-i-Martin. 1991. 'Convergence across States and Regions.' Brookings Papers on Economic Activity 1: 107-58. Brehm, John, and Wendy Rahn. 1997. 'Individual-level Evidence for the Causes and Consequences of Social Capital.' American Journal of Political Science 41(3): 999-1023. Crihfield, J.B., J.F. Giertz, and S. Mehta. 1995. 'Economic Growth in the American States: The End of Convergence?' Quarterly Review of Economics and Finance 35: (Special issue), 551-77. Dunbar-Ortiz, Roxanne. 1995. 'One or Two Things I Know about US: Rethinking the Image and Role of the "Okies."' Queen's Quarterly 102(3): 566-76. Fischer, David H. 1989. Albion's Seed: Four British Folkways in America. New York: Oxford University Press. Helliwell, John F. 1996a. 'Economic Growth and Social Capital in Asia.' NBER Working Paper 5470. Cambridge: National Bureau of Economic Research. - 1996b. 'Convergence and Migration among Provinces.' Canadian Journal of Economics 29 (April): S324-30. - 1996c. 'How Much Do National Borders Matter for Quebec's Trade?' Canadian Journal of Economics 29 (August): 507-22. - 1998. How Much Do National Borders Matter? Washington, DC: Brookings Institution. - and Robert D. Putnam. 1995. 'Economic Growth and Social Capital in Italy.' Eastern Economic Journal 21(3): 295-307. Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton, NJ: Princeton University Press. - Neil Nevitte, and Miguel Basanez. 1995. The North American Trajectory. Berlin and New York: Aldine de Gruyter. Kalin, R., and J.W. Berry. 1996. 'Ethnic and Civic Self-Identity in Canada: Analyses of 1974 and 1991 National Surveys.' Canadian Ethnic Studies. McCallum, John. 1995. 'National Borders Matter: Canada-U.S. Regional Trade Patterns.' American Economic Review 85 (June): 615-23. Platteau, Jean-Philippe. 1994a. 'Behind the Market Stage Where Real Societies Exist: Part I. The Role of Public and Private Order Institutions.' Journal of Development Studies 30(3): 533-77.

42 John F. Helliwell - 1994b. 'Behind the Market Stage Where Real Societies Exist: Part II. The Role of Moral Norms.' Journal of Development Studies 30(4): 753-817. Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. - 1995. Tuning In, Tuning Out: The Strange Disappearance of Social Capital in America.' PS Political Science and Politics 28(4): 644-83. Rice, Tom W., and Jan L. Feldman. 1997. 'Civic Culture and Democracy From Europe to America.' Journal of Politics 59(4): 1143-72. Riddell, W. Craig. 1993. 'Unionization in Canada and the United States: A Tale of Two Countries.' In David Card and Richard B. Freeman, eds., Small Differences that Matter: Labor Markets and Income Maintenance in Canada and the United States. Chicago: University of Chicago Press, 109-47. Verba, Sidney, Kay L. Schlozman, and Henry Brady. 1995. Voice and Equality: Civic Voluntarism in American Politics. Cambridge, Mass.: Harvard University Press.

Social Capital, Social Cohesion, Community: A Microeconomic Analysis1 JEFF DAYTON-JOHNSON

There is a growing concern with the role of trust, social capital, and community, in part because there is so much variation in the indicators of these aspects of societies. In some countries, most people claim to trust each other, while elsewhere most say they do not. The proportion of respondents in a recent survey who agree with the statement that 'most people can be trusted' ranges from 6.7 per cent in Brazil to 61.2 per cent in Norway. People are more likely to join voluntary associations, be they political parties, sports, cultural, or religious organizations, in some societies than others. The same survey reveals that the average number of organizations to which a person belongs ranges from 0.38 in Japan and Italy to 1.5 in the United States and 1.7 in Iceland.2 Some neighbourhoods and regions have a vibrant community life, while others are marked by isolation and decay. Does some combination of a high level of social capital, social cohesion, and community lead to greater well-being and better economic performance? Empirical economic studies argue that it does.3 Two fundamental issues, however, remain poorly understood. First, what is 'social capital' and what is not? Does it differ from 'social cohesion' or 'community'? If so, how? Second, what mechanisms link social capital, social cohesion, and community to economic performance? This chapter uses microeconomic theory to distinguish among social capital, social cohesion, and community, and to clarify the conditions under which these aspects of society influence economic performance. An economic system can be viewed as a network of cooperative behaviour. Market-mediated exchange, even in a competitive setting, is more successful if buyers and sellers can be reasonably assured that they will not be cheated. Sequential exchanges might not happen at all if there is no trust between the parties. The phenomenon is not limited to exchange. Complicated production

44 Jeff Dayton-Johnson processes require coordination among many people, coordination that can be facilitated by trust and cooperation. Furthermore, economic growth is driven by the adoption of ever more efficient technologies and organizational innovations; these adoptions likewise call for coordination, cooperation, and trust. In many of these social interactions, each person faces very strong incentives to exploit another person for personal reward, even though mutual cooperation is better for all parties. This is the quandary of the so-called prisoners' dilemma, the core of the model employed in this chapter. Social capital changes individual incentives towards cooperation and thus allows cooperation to emerge in settings where it otherwise would not. Cooperative behaviour, in turn, is likely to lead to better economic performance for the society (suitably defined) as a whole. This chapter is organized as follows. The next section links the leading ideas regarding social capital with theories of economic growth and development. The basic economic model of social capital is introduced in the section that follows. As is customary in microeconomic modelling, the setting is abstract and stylized, stripping away many of the complicated features of social interactions that exist in the real world. The trade-off is that such an approach yields strong predictions about the outcome of social interactions. In the basic model of social capital, two economic agents must play the prisoners' dilemma game, in which mutual cooperation is socially optimal, but mutual defection is the predicted outcome. The model allows the agents to invest in 'social capital,' defined as a claim on the returns to mutual cooperation, prior to playing the prisoners' dilemma. Social capital as defined here is an individual asset, similar to human capital: an agent must set aside current resources in the hope of a future return. Indeed, in his classic early formulation, Becker (1964) defines human capital investment as 'activities that increase future monetary and psychic income by increasing the resources in people.' In this view, social capital is a subset of human capital. Unlike Becker's view of human capital, however, but similar to the role of human capital in the 'new growth theory' (e.g., Lucas 1988), social capital generates spillovers: if one player invests, it makes such investment more profitable for the other. The social capital model introduced in this chapter provides for conditions under which the opportunity to invest in social capital allows cooperation to emerge in equilibrium. Does the two-person model scale up to a society of many people? To answer this question, a model of social cohesion is presented. This model considers an arbitrarily large society of agents who are randomly matched to play the combination social capital and prisoners' dilemma game (the social capital model) period after period. Agents have no information about the

Social Capital, Social Cohesion, Community 45 history of play of their opponents. In this game, conditions are given under which cooperation emerges in equilibrium. The social cohesion model endogenizes the return to social capital investment as an increasing function of the historical society-wide investment in social capital. Agents know only this aggregate statistic about social capital. The aggregate statistic is called 'social cohesion.' Social cohesion is the (depreciated) stock of past social capital investments. This definition is similar in spirit to the conclusions of the report by the Canadian government's Policy Research Network on Social Cohesion, namely, that 'social cohesion [is] an outcome of investments in social capital.'4 Stanley (1998) elaborates on this view, writing that 'social cohesion is the bonding effect within a society that arises spontaneously from the unforced willingness of individual members of a society to enter into relationships with one another in their efforts to survive and prosper' and that 'social cohesion is strengthened by the existence and creation of social capital.' We then consider how cooperation among agents might be further facilitated if they are members of a society that is, in a very precise sense, a community among a group of randomly matched players like that featured in the social cohesion game of the previous section. A community exists when there is better (though not perfect) information about other agents' actions and where there is a social standard of behaviour. Better information and the social standard of behaviour jointly comprise norms at the community level. The community model shows that under a particular social standard of behaviour with fairly weak information requirements, cooperation can be sustained in equilibrium in a community in conditions where it would be impossible in the social cohesion game (or model). Reference to the various divisions in Canadian and U.S. society, for example, Francophones and Anglophones or ethnic segmentation in major cities, suggests that a society might be made up of segments within which social cohesion is independently generated. Friesen (in her chapter in this volume) calls these segments 'communities.' This type of group is modelled here as marked by random matching of agents within the segment and no matching of agents between segments. If disparate segments share a geographic or political space, even if between-group interactions are rare, then society-wide social cohesion might be determined by aggregate social capital investments across all segments. If this is so, a group with a low initial endowment of social cohesion might nevertheless exhibit cooperative behaviour (borrowing social cohesion, in a sense, from another segment of society). It is also shown that a cooperative minority might be undermined by the uncooperative behaviour of a majority. The concluding section of this chapter introduces several potential extensions of the analysis.

46 Jeff Dayton-Johnson Social Capital in the Small and in the Large To better orient the microeconomic analysis that follows, this section highlights some of the more celebrated perspectives on social capital.5 In the Large Arguably, the most fundamental task of economics is to explain differences in prosperity, that is, development and growth. The average person in a country such as Canada, France, or New Zealand is more than ten times richer than the average person in Vietnam and close to forty times richer than the average person in Mozambique. Why do average incomes differ so vastly between countries?6 Standard models of economic growth identify two sources of economic growth: (1) the accumulation over time of labour and of physical and human capital and (2) improvements in technology and the organization of production. Can economic growth theory, with its focus on the accumulation of productive factors, account for differences on the order of the gap between Canada and Mozambique? To date, the ability of economic growth theory to explain the performance of different economies is mixed at best. For example, Prescott (1998), in his overview of the empirical literature on international differences in income, rejects differences in savings rates (which determine the rate of physical capital accumulation) or investments in 'intangible capital' (i.e., education and training) as being capable of explaining differences of the magnitude of those that separate Canada and Mozambique. Could differences in trust, social capital, social cohesion, and quality of community7 contribute the differences in levels of per capita income? Do differences in measures of social capital correlate with differences in income levels? Knack and Keefer (1997) find that the level of trust - measured as the proportion of survey respondents who say that 'most people can be trusted' is strongly associated with higher per capita income growth in a cross-country regression for twenty-nine market economies. An increase of one standard deviation in the level of trust has an effect on growth similar in magnitude to an increase of one standard deviation in the primary school enrolment ratio. Fukuyama (1995) argues that trust reduces transaction costs and facilitates exchange; countries with a higher level of trust (as a function of religious and cultural factors) have better-functioning firms and more voluntary activity and, therefore, better-performing economies. Helliwell (in this volume) finds that among regions in Canada and the United States, social capital (again, measured by survey responses regarding trust) is highest where income as highest. (Growth of per capita income, however, has been highest in the

Social Capital, Social Cohesion, Community

47

regions with the lowest social capital.) Detecting the effect of social capital on growth is complicated, however, by the effects of convergence: poorer economies tend to grow more quickly than rich economies, conditional upon having achieved certain levels of educational and infrastructure development. Thus, poorer regions of Canada and the United States would likely have grown more quickly than their more wealthy neighbours, regardless of their relative levels of social capital. Vast differences in average incomes among countries pose an unsolved puzzle for economists. Surely, no country would 'choose' an average income level equal to one-fortieth of the Canadian average. The standard economic model of growth signals rates of saving and of investment in education as critical determinants of a country's income level and growth rate; but savings rates and educational levels simply do not vary enough to explain the variation in living standards in the world today. In the search for answers to this puzzle, new research indicates that loosely defined concepts such as social capital, trust, or social capability might contribute to explaining those differences. Often, however, neither the social capital variables themselves, nor the way they work are tightly defined or clearly understood. In the Small What drives changes in total factor productivity? This is the puzzle posed by the vast differences noted above in per capita income. Prescott (1998) cites suggestive studies that show that the adoption of the best productive practices is not automatic, since people can resist their adoption. Even without active resistance, the take-up of innovations is rarely unanimous. As Helliwell points out (in his chapter in this volume), greater associational activity and higher levels of trust might make it easier for economic agents to cooperate and adopt the most productive practices. Case studies from across the social sciences have found social capital to be positively related to changes in productivity. In Putnam's study of society at the local level in Italy (1993b), for example, social capital is defined as the horizontal social relations created by local associational activity. Such activities can enhance economic outcomes if, for example, the performance of government is improved. Indeed, Putnam finds that in Italian regions where people have a greater propensity to join football clubs and choral societies, governments are more efficient (e.g., health care reimbursement claims are more speedily processed), presumably because the local government is more closely monitored by its constituents. Narayan and Pritchett (1999) use data from Tanzania and, like Putnam, compute an index of associational activity.

48

Jeff Dayton-Johnson

They find that an increase of one standard deviation in the value of this index in a village is associated with 20 per cent higher per capita income. Development economists have also been looking at various forms of social capital as important determinants of well-being in poor societies. Some villages have high levels of social capital, in the form of excellent information networks and the availability of social sanctions to enforce cooperative agreements. Even in the absence of insurance markets, poor villagers can form mutual insurance networks that reduce members' exposure to idiosyncratic production shocks. In the absence of credit markets, they can form rotating savings and credit associations to allow small-scale investments. Such arrangements would be impossible without the dense information networks of their community life. In this vein, Ostrom's (1990) study of local regulatory institutions for common property resources - such as inshore fisheries, community forests, or farmer-managed irrigation systems - demonstrates that local community management can be more efficient than formal governmental regulation of the commons and more efficient than establishing a private market in the natural resource. The locally crafted rules and institutions that govern the use of the natural resource constitute an important example of 'community' (according to the definition used in this chapter). This brief discussion of social capital 'in the small,' at the level of individual decision making, suggests that there are important unresolved questions regarding the relationship between social capital and markets. Helliwell and others argue that social capital might make markets (or production processes) work better. Production teams are more likely to adopt more productive techniques, which might require intensive coordination. People will have better channels of information about economic opportunities. Buyers and sellers will be more willing to trust one another. The emphasis of development economics, in contrast, is that social capital facilitates responses to the absence of markets. Mutual insurance networks, local governance for the commons, and rotating savings and credit associations are substitutes for markets that fail to arise for various structural reasons. The issue of whether social capital is a complement to, or substitute for, markets depends on the level of development of an economy. In the context of missing markets, social capital can help substitute for market failure. In more developed economies, where the missing-market problem is not so pervasive, social capital can make markets function more smoothly. Some types of markets, however, may be inimical to social capital. Seabright (1997), for example, argues that privatizing the commons can destroy trust in the resource-using community. Pre-existing relations of cooperation might be

Social Capital, Social Cohesion, Community 49 weakened by the enforcement of private property rights (a necessary precondition for markets), in part because the tradability of property rights undermines reliable long-term relationships among the users of the resource. Measurement Ahead of Theory Current literature contains several definitions of social capital. Putnam's classic framework, like those of Ostrom and the development economists already cited, sees social capital as the network of social relations that link a group of agents: Putnam (1993b) writes of those 'features of social organization, such as networks, norms, and trust, that facilitate coordination and cooperation for mutual benefit.' Coleman maintains that 'social capital inheres in the structure of relations between persons and among persons. It is lodged neither in individuals nor in physical implements of production' (1990: 302). Knack and Keefer (1997) likewise see social capital as a characteristic of the economy as a whole. Bourdieu (1986) echoes these society-wide concerns, while also emphasizing agents' information about each other. According to Bourdieu, social capital is 'the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition - or in other words, to membership in a group - which provides each of its members with the backing of the collectively-owned capital' (1986: 249). Freisen (in this volume) analyses a similar sort of 'community capital,' which is complementary to human capital investment. The World Bank (1999) adopts a kitchen-sink approach and defines social capital as including, but not being limited to, 'social networks and associated norms ... behavior within and among organizations, such as firms ... [and] government, the political regime, the rule of law, the court system, and civil and political liberties.' Emphasis on the formal institutions of governance echoes the important work of economic historian Douglass North (1990). In many ways, in the area of social capital and social cohesion 'measurement' or empirical study is far ahead of 'theory,' especially economic theory. This chapter argues that somewhat different definitions of social capital (and social cohesion and community) may be best suited to different domains of investigation: thus, the study of the well-being of children (as in Phipps's chapter) reasonably emphasizes a slightly different type of social capital than the study of the climate for private investment (as in the chapter by Dick Stanley and Sandra Smeltzer). At this stage of the debate, there is no need to impose hegemonic definitions of these terms. Nevertheless, this chapter does suggest some distinctions in meaning between 'social capital,' 'social cohe-

50 Jeff Dayton-Johnson sion,' and 'community,' in the hope that these distinctions may contribute to researchers' and policy-makers' understanding of the role of these phenomena in economic performance. In summary, the distinctions drawn by this chapter are the following. Social capital is a characteristic of the individual, defined as the current sacrifices in time, effort, or money made by that individual in the hope of promoting productive cooperation or coordination with others. Social cohesion is a characteristic of the society or collective that depends on the history of the accumulation of social capital in the group and which, in turn, affects the incentives for current social capital investments. Community is, similarly, a characteristic of a group and includes reasonably good information about the action of others, as well as social standards of behaviour that promote cooperation and coordination. Oflrrigators and Aircraft Mechanics A practical example of the importance of social cooperation is production in irrigation systems using small water sources and distribution networks that are collectively owned and managed by the farmers who use them. Every year, these farmers must clean the canal network to increase the flow of irrigation water. Canal cleaning, however, creates spillovers for other irrigators. One fanner may choose not to exert tedious effort in cleaning his stretch of the canal, instead free-riding off the effort of his neighbour. But if all farmers think in this way, the equilibrium outcome might be entirely uncleaned canals, and lower flows of water for all.8 This, in turn, creates a private loss for all farmers, because it makes their crop yields lower. What Ostrom (1990) and others have shown is that community investments in rules to manage the commons can transform pay-offs of the prisoners' dilemma game. As a result, mutual cooperation - high levels of canal cleaning, with concomitant high crop yields and incomes - becomes an equilibrium outcome. This is an example of the means by which social capital, social cohesion, and community can affect economic performance. These concerns are not limited to agrarian economies. In technologically sophisticated team production processes, social capital also affects economic outcomes. For example, in servicing airplanes, the care taken by an aircraft mechanic is critically important to the lives of thousands of passengers. Moreover, the impact of the mechanic's due diligence on the safety of passengers depends on the care taken by the electrician, the ground crews at other airports, and a host of other individuals. The management of the firm provides incentives for the mechanic in the form of monitoring and pay, but these go only so far. Some effort is, in practical terms, unobservable. What determines whether the mechanic will 'go the extra mile' when his effort is observable

Social Capital, Social Cohesion, Community 51 Figure 1 The prisoners' dilemma

only to himself? If the mechanic believes that other individuals are exercising great care, he too is likely to be more conscientious, even though no one can observe him, and his pay does not increase because of the extra effort. However, if the mechanic believes that no one else is conscientious, he may ask himself, 'If no one else cares, why should I?' In such settings of complex team production, social cohesion - the knowledge, or at least the expectation, that others are doing their part - improves the productive performance of the economy.9 Social Capital: A Simple Model This section introduces a bare-bones model to illustrate the idea of 'social capital.' Assume that two agents must undertake some kind of transaction, such that each chooses, simultaneously, one of two actions, denoted C ('cooperate') and D ('defect'). Their pay-offs from the transaction are given in Figure 2.1, where b > a > d > c. Player 1 is the 'row' player and player 2 is the 'column' player. Player 1's pay-off is listed first in each cell, player 2's second. This is the prisoners' dilemma, where the unique Nash equilibrium is , while the Pareto optimum is . Now, suppose that each of the players has the option of investing in some social capital prior to playing the game depicted above. Each chooses, simultaneously, whether to invest in social capital, the rate of return to which is r. Afterward, both players observe whether the other invested K - 1 or K = 0,

52 Jeff Dayton-Johnson and they play the game. Social capital is not entirely analogous to physical or human capital: a player only realizes the return to social capital investment if mutual cooperation occurs. In a sense, the player posts a bond, according to which she is penalized if mutual cooperation does not occur. The first substage of the expanded game is the social capital game: the second substage is no longer a prisoners' dilemma if social capital investment occurs in the first substage, and accordingly is now called the cooperation game. There is no discounting of pay-offs between substages. In the analysis that follows, attention is restricted to pure strategies. What is social capital? Consider the example of the small irrigation system. If one farmer attends the annual meeting of the irrigation society, he incurs a cost in time and perhaps tedium. The outcome of the meeting, however, is the coordination of canal cleaning and delivery of irrigation water among all those who attend. If our farmer and his neighbours follow the plan settled upon in the meeting (i.e., they all play C following an investment of K = 1), their pay-offs to cooperation - in the form of privately earned crop yields - are higher than they would have been without the meeting. (The situation is similar for volunteer organizations like those analysed by Woolley in this volume.) Attending a meeting signals a player's willingness to cooperate in the future by raising her pay-off to cooperative behaviour. Her pay-off is raised by learning about the coordination of canal cleaning. If she adheres to the canal-cleaning plan, the return to her cooperative effort will be higher her cooperative effort would have a very low return if instead she opted to clean the canal six months before irrigation was to begin or two days after it began. Furthermore, since her pay-off as well as everyone else's is rising in her canal-cleaning effort, her pay-off increases with her social capital investment. Alternatively, return to the example of the airplane mechanic. Two maintenance workers are players 1 and 2. The cooperation decision is whether to provide a high-effort level (C) or to shirk (D), and part of the effort is unobservable by management. If both shirk, it substantially increases the probability that the aircraft will suffer some kind of mishap, inducing guilt and reducing the long-term income of both workers if the company goes out of business as a result. But suppose, in this simple setting, that both workers have the option of attending a voluntary training session prior to working on the plane. If the first mechanic goes to the training session and sees the second there, it increases his expectation that the second mechanic will not shirk. Formally, each can condition his choice of C or D on whether he saw the other at the session (K = 1) or not (K = 0). Moreover, the training session imparts information that raises both workers' productivity, if both provide high effort subsequently. This, in turn, further reduces the risk of accidents, thereby increasing the firm's success and the lifetime income of the mechan-

Social Capital, Social Cohesion, Community 53 ics.10 Note that although this example might be of special salience for frequent flyers, the issue is really quite general. Unobservable individual effort and large returns to joint effort characterize many job situations (as do opportunities to signal commitment to firms' goals). The first simple result is as follows. Proposition 1. Let b - a K. If this condition is not met, should the uncooperative action be played? Stated another way, if past investment ('cohesion') is less than the critical value K, why would a player deviate from uncooperative behaviour by playing K = 1, or C, or both? If K- KQ is positive but small, one individual j might decide to invest in social capital for a few periods, until such time t that K: _ , >K. At that point, everyone will begin investing in social capital (and cooperating), and player j will earn a return on his investment. Condition (3) guarantees that no one player would ever make such a deviation in a low-cohesion setting. (Proposition 4 in any case demonstrates that such unconditional cooperation can never form a Markov perfect equilibrium of the social cohesion game.) This condition is sufficient to guarantee that the following 'cooperative Markov strategy,' when played by all players, constitutes a Markov perfect equilibrium of the N players social cohesion game.

Social Capital, Social Cohesion, Community 59 Definition 2. Consider the following strategy, called the cooperative Markov strategy. If K> g(b -a- 1), play the following sequence of moves in period t. Play K = 1 in the social capital substage; play C in the cooperation game if is the outcome of the social capital game, and D otherwise. If K c. If j deviates during any period's social capital game, she invests 1. Since all other players play D unconditionally in all periods, there is no way that j can ever realize the return from her social capital investment, and her pay-off is 1 less than it would be by adhering to the uncooperative Markov strategy. Proof of Proposition 4 Suppose that all players play an unconditional cooperation strategy: Regardless of the value of Kt _ , , play K = 1 in the social capital game, and play C in the cooperation game if < 1,1 > is the outcome of the social capital game. (The argument is similar for strategies that prescribe C unconditionally for all outcomes of the social capital game.) Now suppose that £",_,< g(b - a - 1). Then if player j invests K = 1 in period t, but deviates by playing D in the cooperation game, her pay-off in t will be b - 1 rather than a + r. Given that K < t-1 S(b - f l - l X b - \ > a + f ( K t _ , ) = a + r, so/s lifetime pay-off is raised by deviating. Proof of Proposition 5 In the text of the chapter, it is shown that no player could profit by unilaterally deviating from the cooperative Markov strategy when Kt , > K . It remains to be shown that no single player will deviate from the actions pre-

72 Jeff Dayton-Johnson scribed by the cooperative Markov strategy when Kt _ , < K. It remains to be shown that no single player will deviate from the actions prescribed by the cooperative Markov strategy when * " , _ , < £ (namely, ) when Condition (3) is satisfied. The strategy is to define the condition under which the greatest return she could get from deviating is smaller than the smallest possible cost of deviating; it will be shown that this condition is equivalent to Condition (3). As above, the proof need only show that no one-period deviation is profitable (cf. Abreu 1988). What is the greatest possible return she could get from deviating? By deviating this period (and possibly for several periods), she will at some point (say period f) trigger social capital investment by all players and her pay-off from that will be:

In the absence of social capital investment, player /s pay-off would have been 8d + S2d + 83d + . The return from deviating is the difference

The realized values of rt depend on the initial level of social cohesion KTO, the function/(.), and the rate of depreciation of societal social capital p. As an upper bound on rt, consider the maximum possible level to which r =f(K_,) could tend as t —» °°. If there are N players, and every player invests K = 1 in every period, then for arbitrarily large T, K^is given by

As T -» oo,

Then, certainly, r can never be any larger than

so that the most maty could gain from deviating from the cooperative Markov strategy when KO > K is bounded above by

Social Capital, Social Cohesion, Community

73

What is the lowest cost that j would incur in order to trigger social capital investment? Every period that she invests K = 1 rather than playing the prescribed action, her per period pay-off is d - 1 rather than d, so that the per period cost of deviating is just 1. The lowest cost that might possibly exist for such deviations arises when it only takes one period to trigger social capital investment, and that cost is 1. So a sufficient condition such that no single player will deviate from the cooperative Markov strategy is that

which can be rearranged as Condition (3). Thus, if Condition (3) is satisfied, no player will ever deviate from the cooperative Markov strategy when K < K. Proof of Proposition 6 Under what conditions will our players cooperate? First, it must be that no player would wish to deviate from the equilibrium path. If a player cooperates forever, her pay-off is

If she deviates unilaterally by investing K = 0 in one round, then her opponent will play D, and the best she can do is also play d, so that her pay-off would be d. (Note that this is true whether or not her opponent has ever observed deviation in the past.) After deviating for one period, the equilibrium would dictate that she defect ever after, and her pay-off would be d every period. Thus, one condition for the contagious equilibrium is (5)

The proposition assumes that r, > 0. Then (a - d) + rl > 0 and

If all players save one adhere to the contagious equilibrium, then rt > r, > 0, Vt. This, with Condition (6), guarantees Condition (5). It must also be that player j is not tempted to deviate off the equilibrium path, that is, when defection has begun to spread throughout the community. If j has observed defection, then the equilibrium calls upon her to defect also: play K = 0 and D regardless of her opponent's actions. This yields a dis-

74 Jeff Dayton- Johnson counted pay-off of dl(\ - 8). To deviate in this setting is to invest K - 1 in social capital during some round t. Player j might do this for one of two reasons. She hopes to encounter another player who has not yet observed defection, and can either cooperate (and earn a + r) or defect (and earn b - 1). In the latter case, j fools her opponent into thinking she plans to cooperate, only to subsequently take advantage of her. Suppose that the number of players who have observed defection is M. Then the expected value of a oneround deviation forj is

where (N-M)/(N- 1) and (M - l)/(N~ 1) are the probabilities, respectively, of encountering a player who has not witnessed defection or encountering one who has. Player j will not deviate off the equilibrium path as long as

The latter can be expressed as

which is Condition (4) from this proposition. This means that the one-period gain from taking advantage of a cooperative type player (b - 1) is not sufficiently large to entice a defection type of player to try to fool her opponents. Note that Condition (4) is more likely to be met the higher the number of dtype players including our player j; the condition is more binding when j believes that few players other than herself have witnessed defection. REMARK

Alternatively, Condition (4) can be expressed as a (minimum) condition on the size of N or d such that the contagious social standard of behaviour is an equilibrium. Notes 1 My sincere thanks to the editor of this volume, Lars Osberg, for his stimulating comments and warm encouragement. Financial support from the Department of Canadian Heritage is gratefully acknowledged.

Social Capital, Social Cohesion, Community 75 2 Both results are from the World Values Survey, reported in Knack and Keefer's (1997) Data Appendix. 3 See, for example, the studies by Knack and Keefer (1997) and Narayan and Pritchett (1999) discussed in the literature review below. 4 Cited from the Social Cohesion Network's Web site, http:// policyresearch.schoolnet.ca/networks/cohsoc/socialco-e.htm, accessed in July 1999. 5 See also Osberg's introductory chapter. Jenson (1998) provides a critical assessment of the Canadian literature on social cohesion; the University of Maryland's IRIS Center Web site (http://www.inform.umd.edu/IRIS/) includes an exhaustive bibliography on social capital; there, one can access useful working papers, including the bibliographic surveys by Knack (1999) and Feldman and Assaf (1999), both of which were very useful in the preparation of this chapter. 6 Berry, Bourguignon, and Morrisson (1991) demonstrate that, globally, intercountry income inequality clearly dominates mfracountry inequality. 7 Several competing definitions of social capital will be discussed in this chapter. This section considers only the general theme. 8 Strictly speaking, the costs of uncleaned canals, realized in the form of lower output, will not be equally borne, as the farmers at the end of the canal network will suffer the accumulated effects of clogged canals. Nevertheless, all farmers depend on the cleanliness of the main canal that links the water source (reservoir, well, river diversion) to the field channels, and even the first farmer to divert water suffers from the failure of collective canal cleaning. 9 The issue of social relations and productivity in the workplace is formally analysed by Spagnolo (1999). 'Edgeworth externalities' - the return to one agent's effort is increasing in the effort of another - is made the basis of a growth model by Kremer (1993), who illustrates the global importance of this kind of interrelatedness, with applications to human capital accumulation. 10 Given the simple situation laid out in this paragraph, it is in management's interest to offer such training session, but only //"attendance is voluntary. If all employees must attend the training session, there is no scope for signalling (although it may still be the case that the training imparts information). 11 Spagnolo (1999) considers a similar information structure in his model of cooperative relations in production by examining the effect of requiring that all workers in a particular firm come from a single village. This is equivalent in his paper to assuming that all workers can observe the history of social relations. 12 Note that it has been so far assumed that social capital for individuals essentially depreciates at a rate of 100 per cent: each period, a player earns a return from social capital only if she has invested that period. In part, this is a consequence of

76 Jeff Dayton-Johnson

13 14 15 16

17 18

the random-matching mechanism and imperfect information. An important parameter in a more complicated model might be the ratio of the depreciation rates at the individual and aggregate levels. Note also that the 100 per cent rate of depreciation can be viewed as a means for distinguishing social capital from the classic definition of human capital. That is, human capital depreciates more slowly than social capital. This distinction would make an individual's reputation - given that it endures from one interaction to another - a feature of his human capital. This issue has been considered theoretically by Kreps (1990). Cited in PRI (1999); see also SSCSAST (1999). Thus, there are 7+ 1 status levels, 'good,' plus T grades of 'bad,' where Tis the number of periods of punishment. Strictly speaking, it should be noted that the examples given above involving forbearance for irrigators or mechanics with personal problems are far more complicated than the (T+ 1) status social standard of behaviour. This introduces random shocks to players that are not a feature of either the contagious or (T + 1) status models. Cited in Wynn (1987: 265-6). The endogeneity of norms is considered by Bowles and Gintis (1998) in their study of community governance and by Sethi and Somanathan (1996) in their model of governance of the local commons. Both papers use an evolutionary games framework that differs from that of this chapter.

References Abreu, D. 1988. 'On the Theory of Infinitely Repeated Games with Discounting.' Econometrica 56: 383-96. Anderson, B. 1991. Imagined Communities: Reflections on the Origin and Spread of Nationalism. London: Verso. Becker, G.S. 1964. Human Capital. New York: Columbia University Press. Be"nabou, R. 1996. 'Inequality and Growth.' NBER Macroeconomics Annual 11: 11-74. Berry, A., F. Bourguignon, and C. Morrisson. 1991. 'Global Economic Inequality and Its Trends since 1950.' In L. Osberg, ed., Economic Inequality and Poverty: International Perspectives. Armonk, NY and London: M.E. Sharpe, 60-91. Bourdieu, P. 1986. The Forms of Capital.' In J. Richardson, ed., Handbook of Theory and Research for the Sociology of Education. Westport, Conn.: Greenwood Press, 241-58. Bowles, S., and H. Gintis. 1998. 'How Communities Govern: The Structural Basis of Prosocial Norms.' In A. Ben-Ner and L. Putterman, eds., Economics, Values, and Organization. Cambridge: Cambridge University Press, 206-30.

Social Capital, Social Cohesion, Community

77

Coleman, J. 1990. Foundations of Social Theory. Cambridge, Mass.: Harvard University Press. Feldman, T.R., and S. Assaf. 1999. Social Capital: Conceptual Frameworks and Empirical Evidence: An Annotated Bibliography. Social Capital Initiative Working Paper 5. Washington, DC: World Bank. Fudenberg, D., and E. Maskin. 1986. 'The Folk Theorem in Repeated Games with Discounting or with Incomplete Information.' Econometrica 54: 533-56. Fukuyama, F. 1995. Trust: The Social Values and the Creation of Prosperity. New York: Free Press. Jenson, J. 1998. Mapping Social Cohesion: The State of Canadian Research. CPRN Study F03. Ottawa: Canadian Policy Research Networks. Kandori, M. 1992. 'Social Norms and Community Enforcement.' Review of Economic Studies 59: 63-80. Knack, S. 1999. Social Capital, Growth and Poverty: A Survey of Cross-Country Evidence. Social Capital Initiative Working Paper 7. Washington, DC: World Bank. Knack, S., and P. Keefer. 1997. 'Does Social Capital Have an Economic Payoff?' Quarterly Journal of Economics 112: 1251-1332. Kremer, M. 1993. The O-Ring Theory of Economic Development.' Quarterly Journal of Economics 108: 551-75. Kreps, D. M. 1990. 'Corporate Culture and Economic Theory.' In J. Alt and K. Shepsle, eds., Perspectives on Positive Political Economy. New York: Cambridge University Press, 90-143. Lucas, R., Jr. 1988. 'On the Mechanics of Economic Development.' Journal of Monetary Economics 22: 3-42. Narayan, D., and L. Pritchett. 1999. 'Cents and Sociability: Household Income and Social Capital in Rural Tanzania.' Economic Development and Cultural Change 47: 873-97. North, D.C. 1990. Institutions, Institutional Change and Economic Performance. New York: Cambridge University Press. Okuno-Fujiwara, M., and A. Postlewaite. 1995. 'Social Norms in Random Matching Games.' Games and Economic Behavior 9: 79-109. Olson, M. 1965. The Logic of Collective Action. Cambridge, Mass.: Harvard University Press. Ostrom, E. 1990. Governing the Commons: The Evolution of Institutions for Collective Action. New York: Cambridge University Press. Fortes, A., and P. Landolt. 1996. The Downside of Social Capital.' American Prospect 26: 18-21. Prescott, E.G. 1998. 'Needed: A Theory of Total Factor Productivity.' Lawrence R. Klein Lecture 1997. International Economic Review 39: 525-51.

78

Jeff Dayton-Johnson

PRI. 1999. Sustaining Growth, Human Development, and Social Cohesion in a Global World. Ottawa: Social Cohesion Network, Government of Canada, Policy Research Initiative. Putnam, R. 1993a. 'The Prosperous Community - Social Capital and Public Life.' American Prospect. 13: 35-42. Putnam, R., with R. Leonardi and R. Nanetti. 1993b. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Resell, S.A. 1999. Renewing Governance: Government in the Information Age. Don Mills, Ont.: Oxford University Press. Schelling, T.C. 1960. The Strategy of Conflict. Cambridge, Mass.: Harvard University Press. Seabright, P. 1997. Transferability of Collective Property Rights: Does Trade Destroy Trust?' In J. Roemer, ed., Property Rights, Incentives and Welfare. London: lAE/Macmillan, 94-111. Sethi, R., and E. Somanathan. 1996. The Evolution of Social Norms in Common Property Resource Use.' American Economic Review 86: 766-88. Spagnolo, G. 1999. 'Social Relations and Cooperation in Organizations.' Journal of Economic Behavior and Organization 38: 1-26. SSCSAST. 1999. Final Report on Social Cohesion. Ottawa: Government of Canada. Standing Senate Committee on Social Affairs, Science and Technology. Stanley, D. 1998. Hearing Secret Harmonies: Toward a Dynamic Model of Social Cohesion. Mimeo. Hull, Que.: Department of Canadian Heritage. Tamura, R. 1991. 'Income Convergence in an Endogenous Growth Model.' Journal of Political Economy 99: 522-40. World Bank. 1999. Social capital web site, www.worldbank.org/poverty/scapital. Wynn, G. 1987. 'On the Margins of Empire (1760-1840).' In C. Brown, ed., The Illustrated History of Canada. Toronto: Key Porter Books.

Social Cohesion and the Well-being of Canadian Children SHELLEY PHIPPS1

This chapter explores some potential links between social cohesion and children's well-being in two main stages: (1) a conceptual discussion of how social cohesion might affect the well-being of children and (2) two illustrative empirical tests of the hypothesis that higher levels of social cohesion generate higher levels of well-being for children. First, the argument is made for a child-centred approach to understanding the well-being of children, and some existing economic definitions of well-being are evaluated in terms of their suitability for children. Discussion of a child-centred approach to social cohesion follows. Then, some economic models are surveyed of the determinants of children's well-being, and Coleman's (1988) idea of social capital is discussed. The first part of this chapter ends with a summary of an approach to understanding children's well-being in which social relations play a pivotal role. The second part of the chapter consists of two parts. First, microdata from the 1995 Statistics Canada General Social Survey (GSS) are used to conduct an empirical investigation of the impact of parental relationships on children's well-being from an investment perspective, that is, after the children become adults. Second, microdata from the 1994 Statistics Canada National Longitudinal Survey of Children and Youth (NLSCY) are used to study the links between social support for parents, neighbourhood social support and residential moves, and children's well-being today. The conclusion forms the final part of the chapter. Where Have All the Children Gone? If we are to understand the well-being of children, we must keep children not their parents - at centre-stage. Yet, as argued by Phipps (1999), children, as children, seldom appear in economic models. Although economists have

80 Shelley Phipps addressed the impact of children on consumer behaviour (e.g., Browning 1992), on the labour supply of mothers (e.g., Nakamura and Nakamura 1992), or on the income needs of families (e.g., Deaton and Muellbauer 1986), for example, the focus of these models is on how children affect adult behaviours rather than on the children themselves. There is also an economics literature dealing with the determinants of children's attainments (e.g., the excellent survey by Havemen and Wolfe 1995).2 However, in this research, children are, again, not present as persons now. Children today are essentially treated as consumer durables chosen by their parents at the expense of other forms of consumption (e.g., Becker 1991: chapters 5 and 6; Becker and Tomes 1979, 1986). In this literature, the question of interest is taken to be what happens to children after they become adults, and an investment perspective is taken - if we do not take care of children now we will generate problems for them (and us) later in life as they are less productive economically, and so on. As argued by Jens Qvortrup (1990: 8), 'Children are reduced to "human becomings.'" Of course, becoming a 'successful' adult is very important. We do care about what happens to our children after they grow up. But children are people now, too. Thus, it is important to use a definition of well-being that can pay attention to both present and future. This means that several standard measures of economic well-being will not be appropriate because they are not suitable for application to children. Children are not simply miniature versions of their parents. They differ from adults in terms of needs, opportunities, and capacities (e.g., children are extremely dependent on the significant adults in their lives; they have no personal access to income; they are typically still in the process of developing cognitive and linguistic capacities; they have more limited life experiences; they have limited agency; and their discount rates are higher). With adults in mind, many economists have conceived of well-being as 'utility,' where utility is derived from the consumption of goods and/or leisure time or from income, depending on the theoretical model employed. Phipps (1999) argues that this approach does not work well for children because: (1) it assumes that preferences are given, yet children are clearly in the process of forming and/or changing their preferences as they develop; and (2) observed behaviour is taken to represent utility-maximizing behaviour, yet children have very few opportunities to make their own choices (e.g., family expenditure or income data more likely reflect parental choices). Equally, using a household rather than an individual utility function is inappropriate if we want a child-centred approach because this ignores the possibility of inequality within families (e.g., some children may be abused or neglected).

Social Cohesion and the Well-being of Canadian Children

81

A second approach to assessing economic well-being is to employ selfevaluations (see Hagenaars 1986 for a subjective approach to measuring poverty). Phipps (1999) also argues that this approach has limited applicability for children. First, there is the conceptual problem that a 'spoiled child' may feel unhappy despite having, objectively, a great deal. Adults might distinguish between a selfish and moral self (e.g., they may discount their own feelings of discontent knowing that they are actually better off than many others), but this is not possible for young children who have not yet developed a moral self. And, there are practical problems with a subjective approach to measuring the well-being of children. Although older children may be able to provide self-evaluations of well-being, younger children and infants will not have the linguistic or cognitive skills to make such assessments. We would need, instead, to rely on the assessments of parents (however, Curtis et al. 2000 demonstrate that parents and children do not make the same assessments of a child's well-being). In practice, economists are most likely to proxy well-being through a measure of income.3 This might be regarded as an input orientation which is much narrower, but still consistent with Rawls's (1971) suggestion that a person's level of well-being be measured in terms of 'primary social goods.'4 A focus on the availability of inputs that could generate well-being avoids the self-evaluation difficulties discussed above, but raises another potential problem - children (or adults) may differ in their needs - that is, their ability to process inputs into the realized output of well-being. For example, a child who must use a wheelchair to get around will, compared with others who do not need to purchase wheelchairs, be less well-off if living with the same income (ignoring any associated discomfort, inconvenience, restrictions, or stress). In addition to this general processing problem, the deficiencies of the economists' approach of focusing solely on money income as a measure of wellbeing are particularly large for children because unpaid activities carried on within the home (e.g., reading stories or cooking dinner) have no impact on monetary well-being. Moreover, because income information is most often only available at the household level, researchers must usually assume that income is shared equally among all household members, even though children have no personal control over (family) income (see Phipps and Burton 1995). Overall, Phipps (1999) argues that the approach best suited to a childcentred measure of well-being is Sen's (1992) idea of 'functionings.' Examples of elementary functionings are: being adequately nourished, being in good health, and avoiding escapable morbidity or premature mortality. An example of a more complex functioning is having freedom from fear and

82 Shelley Phipps anxiety. One conceptual advantage of this approach is that functionings, basic or complex, are characteristics of the child, whereas income, for example, is not. Moreover, if we believe that social relationships are central to the wellbeing of the child, then an important advantage of Sen's approach is that the list of functionings can directly include loving or caring dimensions such as being attached to family or having friends.5 Each of the functionings discussed so far make sense whether we are talking about an adult or a child. Thus, they are relevant whether we are talking about the well-being of a child now, or whether we are interested in the well-being of children once they become adults. For most people, however, the capacity to exercise functionings will expand over time. For example, 'taking part in the life of the community' makes more sense at 15 or 50 years than at 5 months. These ideas are developed further in the summary section at the end of this part of the chapter. A Child-Centred Approach to Social Relationships If we are to understand the implications of social cohesion for the well-being of children, some adjustments must be made to our understanding of social cohesion.6 One definition provided by Stanley (1997: 1) is: 'Social cohesion is about all of us "sticking together": it is the bonding effect of that web of social relationships through which individuals are attached to each other in a society, and through which they help each other, knowingly or inadvertently, to achieve their full potential.' For children, rather than focusing on a web of social relationships, we might imagine social relationships in ever-larger concentric circles around the child - imagine the child as the smallest 'Russian doll.' The child will be directly affected by relationships closest to him or her, and more indirectly affected by remote relationships. For example, a child will be affected by relationships within the immediate family; with extended family; with family friends; with the child's own-age friends; with day care providers, teachers, and coaches; by relationships within religious or ethnic groups; by relationships in the community or municipality (e.g., Are there soccer teams, swimming lessons, and Brownies?); and by relationships at the provincial and national level (What is the quality of health and education programs?). Economists interested in explaining outcomes for children have paid relatively little attention to social relationships as a potential contributing factor. This contrasts with work in psychology or sociology, for example. Scholars such as Bronfenbrenner (1979), Cochran et al. (1990), and Garbarino (1979)7

Social Cohesion and the Well-being of Canadian Children

83

have argued that social relationships are key to children's development. For example, Gabarino (1979: 360-1) writes: 'The need for interpersonal support is the fundamental human need, satisfaction of the need is the foundation of social quality ... For the child to flourish, the parent must have access to the social riches of family or family surrogates, kin and kith.' Good relationships within smaller social groups will affect the well-being of the child directly. Loving relationships with parents and grandparents, for example, are central to a child's world. Parents' friends may also have a direct influence on a child's life, for example, as additional adult role models or confidants. To a very young child, parents' social relationships will largely predetermine the child's set of social relationships outside the family (though children participating in day care outside the home will have their own world of relationships with peers and caregivers). As the child ages, he or she will be directly influenced by ever-larger circles of people through participation in more activities and as peer groups become increasingly important to happiness, self-esteem, and behaviour. Of course, there may be differences in the number and intensity of the relationships affecting the child, depending on his or her personality and tastes, and some would argue (e.g., Cochran et al. 1990) possibly on the sex of the child. The impacts of social relationships on children's well-being discussed so far have been direct8 - good relationships are actually components of wellbeing. However, social relationships can also affect well-being indirectly. For example, parents with strong social support may feel less frustrated and/or depressed and, hence, be less irritable or impatient with their children. Adults who are strangers to the child but feel a connection to the community and are willing to volunteer, for example, will make soccer teams and Brownie groups available for all local children.9 Voter willingness to support excellence in health and education for all children will improve children's well-being. In both of these last examples, the social relationships that are important may not actually include the particular child, but the child is affected by the outcome nevertheless (e.g., the availability of the soccer team). Figure 1 illustrates, from a child's perspective, a possible set of social relationships with the child nested in the centre. Of course, for each child, differences in family circumstances and personality will mean that some relationships will be of greater or lesser importance than indicated in Figure 1. For example, for one child, an aunt, uncle, and cousins may live at a great distance and, from the child's point of view, may not really exist outside the family photo album. For another child, an aunt may be a daily caregiver and a cousin a very best friend. To emphasize potential differences in individual

84 Shelley Phipps Figure 1 A child-centred view of social relationships

circumstances, and thus the social relationships that constitute and affect the child's well-being, we also present Figure 2. In contrast with the rich set of social relationships depicted in Figure 1, the child in Figure 2 is very alone in the world, perhaps because the mother, father, and child moved to a new place with brighter job prospects and then the mother and father separated. Other more remote relationships are left in the diagram because, while the child's life may be rather empty of personal relationships, he or she can still potentially benefit indirectly from good social relationships among others (e.g., if the neighbours organize a cub group or if the voters support excellence in local schools). While the idea of the 'Russian doll' is useful in thinking about all the different kinds of social relationships affecting children's well-being, the metaphor is clearly too simple because outside layers cannot always be removed without affecting inside layers. Norms and expectations in the community, for example, might influence parenting behaviour and so relations between

Social Cohesion and the Well-being of Canadian Children 85 Figure 2

A child-centred view of social relationships

parent and child.10 Changes in voter support for the allocation of counselling support through the health care system could affect levels of depression in some parents and hence their direct relationships with their children. Economic Models of the Determinants of the Well-being of Children: Is There a Role for Social Relationships? Existing economic models of the determinants of children's well-being have not typically incorporated a role for social relationships. Most of the economics literature uses conceptions of well-being that are considerably narrower than the functionings approach described in the previous section (e.g., the child's income, educational level, and labour market experiences when he or she becomes an adult) As well, the investment perspective rather than that of the well-being of the child now generally dominates the economics discourse. Becker (1991: chapters 5 and 6) and Becker and Tomes (1979; 1986), for example, assume that children's well-being once they become adults essentially depends on the investment decisions made by their parents. The focus of their analysis is on parental decision making," and children today appear nowhere in the model: 'Children are born in and accumulate human and nonhuman capital in the rth generation and work, consume, and produce their own children in the t + 1 generation' (Becker and Tomes 1979). Becker assumes that parents allocate resources between personal consumption today

86 Shelley Phipps and investment in the future of their children in order to maximize their (own) utility today. Utility maximization occurs subject to the constraint of available income and the relative prices of consumer goods versus investment in children. Children's well-being tomorrow will depend on how much parents choose to invest in them today (as well as upon the genetic and possibly material asset endowments that they may have inherited from their parents and on any 'pure luck' they may experience). Investing in children means making 'expenditures on their skills, health, learning, motivation, "credentials," and many other characteristics' (Becker and Tomes 1986: S5). The prediction of this framework is that children's incomes will depend on parents' incomes (positively) and the number of other children in the family (negatively, because additional children mean less money and time to spend on any one child). Leibowitz (1974) builds on this basic framework by adding the idea that investments in children depend on the amount and quality of time parents spend with them as well as on material investments. Quality of time with children is assumed to increase with the educational level of the parents. Thus, parental choices about, for example, labour supply, will determine both how much money and how much time is available for children. No mention is made of the relationship with the parent, though if there is a connection between spending more time together and having better relationships, then it would be possible to expand the model to allow for the possibility of social relationships mattering for children's attainment in their later life. However, this is not explicit in the Leibowitz model. Moreover, the focus is still on outcomes for children once they become adults rather than on their wellbeing today. Haveman and Wolfe (1995) expand this basic framework to include social choices (e.g., in the form of institutions and programs) and children's choices (e.g., with respect to effort devoted to education) in addition to parental choices. The outcomes discussed are the same (i.e., income and education of the children once they become adults), but the introduction of the idea of social choices allows for the possibility that social cohesion plays a role in the determination of children's outcomes at the level of the state. For example, a more cohesive society might be more prepared to support excellence in educational programs. (However, this is not a point developed by Haveman and Wolfe.) The basic models of determinants of child outcomes commonly found in the economics literature have been expanded by Coleman (1988) to include social capital. Social capital exists when relationships among persons function as resources that can be used to 'facilitate action' or to 'achieve the

Social Cohesion and the Well-being of Canadian Children 87 interests' of the persons involved (1988: SI00-1). According to Coleman, social capital helps to create human capital in the next generation. He highlights roles both for social capital within the family and for social capital outside the family. Within the family, social capital exists in the relations between children and parents (1988: SI 10). Coleman's example is of highly educated parents who spend no time with their children (or who do not get along well with their children even if they are physically present). In such circumstances, the high levels of human capital possessed by the parents can be of little benefit to the children - little new human capital will be produced. Social capital outside the family is defined by Coleman (1988: SI 13) to exist in the relations among parents (of different children) and in the relations of parents with institutions of the community. Coleman argues that social capital is likely to be greatest in situations where parents interact with other parents in a variety of different settings (e.g., at school meetings, at the workplace, in social clubs, and/or at church). Garbarino (1979) and Cochran et al. (1990) make similar arguments about the importance of the parents' social support and social networks, respectively, for the development of the child. As defined by Coleman, social capital is an asset of an individual, not a characteristic of society. Summary of a Child-Centred Definition of Well-being Although much of economic theory essentially ignores children except insofar as they affect the utility of their parents, a child-centred approach suggests that children should be viewed as junior citizens, that is, members of the collectivity in their own right, rather than simply as consumer durables of their parents. This is not to say that the well-being of children once they become adults is unimportant - it is simply to argue that social well-being today depends on the well-being of all people who are members of society today. To understand the well-being of children while children, a useful approach is the 'functionings web' depicted in Figure 3. Here, well-being consists of a multidimensional, mutually interdependent set of beings, doings, havings (e.g., health, good relationships, and security). As already argued, Sen's functionings approach is particularly well suited for application to children. First, the functionings approach depicts a more attractive conception of what it means to be human than the mechanistic U(Y) - 'utility factory' image, for example. Second, it allows for a child-centred approach insofar as direct experiences of children themselves can be incorporated (e.g., health and freedom from anxiety). Finally, social relationships are incorporated as components of well-

88 Shelley Phipps Figure 3 A child-centred definition of economic well-being

being (being attached to family or having friends). This seems essential for children, for whom social relationships are central.12 More indirectly, other social relationships can affect the most primary relationships, from the perspective of the child (e.g., the parent-child relationship). Cochran et al. (1990) suggest that social support can, for example, enhance the parents' own development (e.g., a depressed parent may be less responsive to a child's needs) and can affect child-rearing attitudes and behaviour directly through explicit encouragement or criticism and/or indirectly through modelling and group norms. These are very likely to affect the parent-child relationship.13 In fact, social relationships can affect almost any of the functionings listed above (e.g., health, education, or freedom from anxiety). For example, at the community or municipal level, the degree of civic involvement by adults will affect children's opportunities to learn, to develop friendships, to improve health and fitness, and to participate in the community through the availability of recreational activities and clubs. Functionings14 are represented in a web to emphasize the important connections among components of well-being. High levels in some dimensions may generate the strength and/or resiliency required to withstand difficulties in another. For example, a child with enough to eat is better equipped to understand lessons at school and thus to receive a good education; a child

Social Cohesion and the Well-being of Canadian Children

89

with strong family relationships may be better able to cope with health problems. On the other side, the trauma of parental divorce may have negative consequences for a child's physical and/or emotional health. Deprivation at any point in time may have long-term consequences in terms of reduced capacity to function. Similarly, richness at any point in time may improve future resiliency. Dynamic linkages are thus very important for a full understanding of children's well-being. How does all of this relate to more standard definitions of well-being? Utility is commonly defined over either income or consumption. While income is, of course, vital for the purchase of goods and services that benefit the child, it is not clear which portion of family income is actually being used to benefit the child. In terms of consumption, the child's private goods and services (e.g., food, toys, or swimming lessons) as well as his or her consumption of family-level public goods (e.g., housing, heating, and transportation) will all be important components of well-being. As well, the child will consume goods and services provided by the state (e.g., health and education) and by private institutions (e.g., churches). All of these are inputs to wellbeing, but they are not really well-being in themselves. To this fairly standard list of inputs is added the idea that indirect social relationships can also be inputs to well-being. That is, for example, relationships among individuals unknown to the child can generate services (such as soccer leagues and cub groups) that are direct inputs to the child's well-being. If we want to understand children's well-being while they are children, we can use the idea of a functionings web. This has the advantage that it can also be used to understand the well-being of adults, particularly if we allow the functionings web to change and expand over time with expanded life experiences (see Figure 4). This conception, hopefully, loses none of the insights of earlier work (e.g., income is an essential input to the well-being of children or adults), but adds to the story an emphasis on the importance of social relationships. The remaining part of this chapter goes on to conduct preliminary empirical tests using Canadian data of two aspects of the hypothesis that social relationships matter for children's well-being. Long-Term Consequences of Childhood Relationships with Parents: An Empirical Study Using the Investment Perspective Coleman (1988) argues that poor relationships with parents can impede children's development and thus have potential negative consequences for

90

Shelley Phipps

Figure 4 Dynamic linkages & expansion of the functionings web over time

their well-being as adults (as well as during childhood). The 1995 Statistics Canada General Social Survey (GSS) asks respondents for retrospective information about childhood relationships with parents as well as asking them for current (i.e., adult) information. This section provides an econometric investigation of the links between adult well-being as reflected through a set of current functionings and indicators of childhood relationships with parents. We study adults aged 25 to 54 at the time of the survey, most of whom should be finished schooling, but not yet be retired. The set of functionings studied includes both having components (e.g., income, health, and happiness) and caring components (e.g., quality of current relationships with spouse, children, or parents, as applicable). Income is pretax total household income from all sources.15 The mean for male respondents is $45,977; the mean for female respondents is $24,567. The second component of well-being studied is satisfaction with current main activity (working for pay, keeping house, looking for a job, going to school, being retired, or being on long-term disability). Only two categories of response were available; over 80 per cent of both men and women reported themselves as satisfied with this dimension of well-being. Having good health is a third important functioning. The question asking about health status compared with others of the same age offers respondents a choice of five categories (excellent, very good, good, fair, and poor). Roughly one-third of respondents chose 'excellent'; less than 10 per cent indicated fair or poor health. (See Table 1.)

Social Cohesion and the Well-being of Canadian Children

91

Table 1 Means/frequencies of 'functionings,' 1995 General Social Survey Males 25-54 Females 25-54 Income (mean value in dollars) Satisfied with main activity

Yes No Looking for job

Respondents main activity in the past 12 months Health status compared to Excellent Very good others of same age Good Fair Poor Strongly agree Having children made respondent a happier Agree person Disagree Strongly disagree Relationship with spouse/ Very happy Fairly happy partner is ... Not too happy Frequency of contact with Daily mother (past 12 months) At least once per week At least once per month Less than once per month Not at all Frequency of contact with Daily father (past 1 2 months) At least once per week At least once per month Less than once per month Not at all

45,976.68

24,567.15

84.8 15.2

87.5 12.5

2.8

1.7

31.0 36.7 22.7

32.2 36.8 22.4

7.6 2.1

6.4 2.1

46.3 49.6

46.8 48.4

3.6 0.5

4.3 0.6

75.3 22.6

75.6 22.0

2.1 6.0

2.4 8.6

34.3 25.5 25.8

37.0 21.2 27.5

8.4 7.4

5.6 6.7

31.5 26.3 26.7

34.9 20.6 29.4

8.1

8.4

Four relationship variables are also considered. Relationships with the respondent's children were assessed using the question: 'Do you agree or disagree with the following statement: Having children made you a happier person.'16 (This question was, obviously, only asked of individuals with children.) Less than 5 per cent of the sample disagreed with this statement, though answering 'agreed' rather than 'strongly agreed' may indicate some problems (roughly equal numbers fell in each category for both male and female respondents). Respondents were also asked about their relationship with their spouse or partner (when relevant). Three categories of response were available (very happy, fairly happy, and not too happy). Only about 2 per cent of either men or women chose 'not too happy.' Finally, we are interested in the respondent's current relationship with his or her parents, if

92 Shelley Phipps Table 2 Frequencies of social cohesion variables for females 25-54, 1995 General Social Survey Strongly agree

Agree

Disagree

Strongly disagree

Close to Mom

31.4

53.1

13.2

2.3

Close to Dad

20.2

48.9

24.1

6.7

Number of observations

2072

Table 3 Frequencies of social cohesion variables for males 25-54, 1995 General Social Survey Strongly agree

Agree

Close to Mom

29.3

62.1

7.3

1.2

Close to Dad

15.0

53.5

26.4

5.1

Number of observations

Disagree

Strongly disagree

1903

parent(s) are still alive. Because no questions are directly asked about the quality of these relationships, we choose the proxy of 'frequency of contact.'17 Although limited by the information available in the GSS, this list of functionings seems reasonable - notice that it contains both traditional economic measures of well-being (e.g., income) and some of the broader components argued for earlier (e.g., happiness and relationships). Table 1 reports means and frequencies for all functionings. The goal of this section of the chapter is to assess the importance of social relationships during childhood for the above dimensions of well-being for adults. The GSS measure of quality of relationship with parents during childhood is both a subjective and a retrospective one, asking respondents whether they were 'very close emotionally with father/mother when growing up (before the age of 15).' Because men and women may respond differently to the social relationships in their lives, we have divided our sample by sex. Of the 2,072 women in the sample, 84.5 per cent agreed or strongly agreed that they were close to their mothers, while 69.1 per cent agreed or strongly agreed to being close to their fathers (see Table 2). Of the 1,903 men in the sample, 91.4 per cent

Social Cohesion and the Well-being of Canadian Children 93 agreed or strongly agreed with the statement: 'I was very close emotionally with my mother when growing up,' while only 68.5 per cent agreed or strongly agreed that they had been emotionally close to their fathers (see Table 3). Thus, both men and women are more likely to report themselves as having been close to their mother than to report themselves as having been close to their father. Men are slightly more likely to report themselves close to their mothers than are women. On the basis of earlier discussion in this chapter, we hypothesize that children experiencing poor relationships with their parents during childhood will have, other things being equal, lower levels of well-being as adults. To test this hypothesis, we estimate a series of equations (probit if the response variable is bivariate, ordered probit if the response variable has several logically ordered categories, tobit for income which can take a zero value for some respondents). In addition to the social relationship variables, we include a set of standard sociodemographic characteristics: current age of respondent, number of children currently in the household, current level of education, current marital status, and current household income (except that income is, of course, excluded for the income equation). Because it is not possible to compare estimated coefficients across equations with different dependent variables and estimated using different econometric techniques, we summarize results with respect to the relationship with Mother and Father variables in Tables 4 and 5.18 For women, childhood relationships with both parents appear to affect current well-being. Relationship with Father is statistically significant in 5 of 8 cases, with the expected positive sign each time. A good relationship with Father during childhood is associated with higher current personal income, increased satisfaction with main activity, and better relationships with children, spouse, and father (currently). Relationship with Mother is statistically significant in 3 of 8 cases, with the expected positive sign in each case. A good relationship with Mother during childhood is associated with better health, increased satisfaction with current main activity, and a better relationship with Mother now. For men, relationship with Father during childhood is a very important predictor of current well-being - the variable is statistically significant for all eight outcome categories, and always with the expected positive sign. Relationship with Mother is only statistically significant in one case, and then with a negative sign. (Having had a close relationship with mother during childhood is associated with lower income currently.)19 However, statistical significance does not necessarily mean big effects. Table 6 uses estimated coefficients from the ordered probit model for current health to predict the size of the association between having had a close

Table 4 The effect of social relationships in childhood on current indicators of well-being for females 25-54, 1995 General Social Survey

Frequency of contact with Dad

Frequency of contact with Mom

Health status

Satisfaction with main activity

n/s n/s All

n/s +

+ +

+ n/s

+ n/s

+ n/s

n/s +

All

All

2072 Probit

2072 Ordered Probit

2072 Probit

All with spouse 1384 Ordered Probit

All with kids ever 1575 Ordered Probit

All with father 1255 Ordered Probit

All with mother 1646 Ordered Probit

Personal income Agree close to Dad Agree Close to Mom Sample description

+ n/s All

n

2072 Tobit

Estimation technique

Having children made respondent happier

Relationship with spouse/ partner

Unemployed during past 12 months

Note: + indicates coefficient is positive and significant; - indicates coefficient is negative and significant; other control variables for these regressions include: age cohorts, number of kids in household, number of kids ever, level of educational attainment, and marital status.

Table 5 The effect of social relationships in childhood on current indicators of well-being for males 25-54, 1995 General Social Survey Unemployed during past 1 2 months

Health status

All

_ n/s All

+ n/s All

1903 Tobit

1903 Probit

1903 Ordered probit

Personal income Agree close to Dad Agree close to Mom Sample description

n Estimation technique

+

Having children made respondent happier

Frequency of contact with Dad

Satisfaction with main activity

Relationship with spouse/ partner

+

+

n/s All

n/s

+ n/s

+ n/s

All with spouse 1362 Ordered probit

All with kids ever 1327 Ordered probit

All with father 1135 Ordered probit

1903 Probit

Frequency of contact with Mom +

n/s All with mother 1516 Ordered probit

Note: + indicates coefficient is positive and significant; - indicates coefficient is negative and significant; other control variables for these regressions include: age cohorts, number of kids in household, number of kids ever, level of educational attainment, and marital status.

Table 6 A comparison of the predicted effects of 'cohesion' variables with effects of education, 1995 General Social Survey Females Outcome question:

Base case*

With postsecondary credentials

Disagree Less than high school close to education Dad

Probability

Point change from base

Point change from base

Point change from base

-0.09461 -0.01142 0.05439 0.03420 0.01744

n/s n/s n/s n/s n/s

Males Base case*

With postsecondary credentials

Less than Disagree high school close to education Dad

Disagree close to Mom

Point change from base

Probability

Point change from base

Point change from base

Point change from base

Point change from base

-0.06648 -0.00488 0.03806 0.02244 0.01087

0.3194 0.37920 0.22082 0.06586 0.01464

n/s n/s n/s n/s n/s

-0.09844 -0.01276 0.05650 0.03929 0.01541

-0.03539 -0.00089 0.02000 0.01207 0.00421

n/s n/s n/s n/s n/s

Disagree close to Mom

In general would you say your health is: excellent 0.32257 very good 0.37198 good 0.22478 fair 0.06208 poor 0.01859

0.06997 -0.00754 -0.03785 -0.01754 -0.00704

*Base case: The base case is an individual who is 35-44 years of age who is married with children present in the household and has a high school diploma. He/she is close to his/her mother and father.

Social Cohesion and the Well-being of Canadian Children

97

relationship with Mother or Father during childhood and current health status. And, to put this in perspective, Table 6 also presents effects associated with having a high or low education. That is, we ask, relative to a base individual who had a good relationship with Father, by how much does the probability of having excellent health fall for an individual who had a bad relationship. Similarly, relative to the base individual who had high school education, by how much does the probability of having excellent health fall for an individual who had lower education. In each case, we compare probabilities with the same base case. Results are also depicted in bar graph form for the probability of having excellent health in Figures 5 and 6. (If a variable was not found to be statistically significant, we do not perform the calculation.) Figure 5 indicates that the quantitative importance of the relationship variables is not trivial, in comparison with the effects of education. For example, for women, having had a bad relationship with Mother during childhood results in a reduction in the probability of having excellent health currently which is very similar to the effect of having low education. For men, the effects of low education on health are larger, while the effect of a bad relationship with Mother is smaller. Still, the social relationship variable is nontrivial by comparison with another more traditional variable, namely, education. The bottom line is that the quality of early childhood relationships matters - a lot - to eventual adult outcomes. Social Cohesion and the Current Well-being of Children: An Empirical Study Using the Perspective that the Child Counts Now This section explores the idea that the well-being of children today depends on the social support available to their parents and/or on the strength of social relationships in their neighbourhoods. In keeping with the arguments made earlier in this chapter, the well-being of the child is understood in terms of a set of functionings that includes both having and loving (caring) elements. The data set employed is the 1994 Statistics Canada National Longitudinal Survey of Children and Youth (NLSCY), a nationally representative sample of children aged 0 to 11 years in 1994. The information about the children used in this analysis was provided by the person most knowledgeable about the child (PMK) - the mother in most cases. Individual functionings are used as dependent variables. Control variables include standard sociodemographic information (e.g., family structure, Mother's education, low-income status, sex and age of child, and number of siblings),20 but the focus here is on adding measures of social relationships.

98 Shelley Phipps Figure 5 The quantitative importance of childhood relationship with mother for present health status. Women currently aged 25-54.

Source: 1995 General Social Survey, Cycle 10: The Family. Statistics Canada Housing, Family, and Social Statistics Division

We first examine the hypothesis that social connectedness of the parent is an important predictor of the well-being experienced by the child, controlling for other relevant factors. In the NLSCY, PMKs are asked six questions to gauge social support; these are then used to construct an index of social support.21 (See Appendix for details.) As a sensitivity check, we estimate our models with social support measured in three different ways: (1) using the social support index directly; (2) including a dummy variable to indicate low social support (i.e., having social support less than 50 per cent of median); (3) including responses to the three social support questions separately. We also test the importance of social support in the neighbourhood for children's current well-being. Neighbourhood may be particularly important for children who often spend more time interacting with neighbours, playing on the street, in the parks, or in other people's backyards, as relevant. The neighbourhood index is constructed from five questions including: 'People around here are willing to help their neighbours.' 'You can count on adults in the neighbourhood to watch out that children are safe and don't get in trouble'

Social Cohesion and the Well-being of Canadian Children 99 Figure 6 The quantatative importance of childhood relationship with mother for present health status. Men currently aged 25-54.

Source: 1995 General Social Survey, Cycle 10: The Family. Statistics Canada Housing, Family, and Social Statistics Division

(see Appendix). We again test the sensitivity of our results by entering the index directly and by including a low neighbourhood score dummy variable (= 1 if the neighbourhood score is less than 50 per cent of median). Finally, since Coleman (1988) has argued that the number of times a child changes schools is a good proxy for potential social capital outside the family - because frequent moves mean less opportunity for the development of connections in the community - we include a dummy variable for having changed residence. Table 7 presents means and frequencies of the social relationship variables. For the social support index, the mean value reported (for 0- to 11-year-olds) was 14.5 relative to a maximum possible score of 18 (81 per cent). Only 1.2 per cent of PMKs reported social support less than 50 per cent of the median (15). For the neighbourhood index, the mean reported value was 10.6 relative to a maximum possible score of 15 (71 per cent). Only 1.9 per cent of PMKs reported very low scores (less than 50 per cent of the median score of 10). Percentages of PMKs (recall that PMKs are nearly always women) reporting involvement in voluntary activities increases with the age of the child, with nearly half (48 per cent) of PMKs with 4- to 11-year-old children volunteering. Frequency of residence change also increases with the age of the child, with 62 per cent of children age 4 to 11 having changed residence at least once.

100 Shelley Phipps Table 7 Means/frequencies of 'social cohesion' variables, 1994 National Longitudinal Survey of Children and Youth Ages 0-11 Ages 2-11 Ages 4-11 Mean value of Social Support Index (Maximum possible score = 1 8 indicating high support) Mean value of Neighbourhood Index (Maximum possible score = 15 indicating high cohesiveness) Dummy = 1 if low social support Dummy = 1 if low neighbourhood score Dummy = 1 if there is someone to trust when having problems Dummy = 1 if there are people to count on in an emergency Dummy = 1 if no one would help if something went wrong Dummy = 1 if there is any parental involvement in voluntary organizations Dummy = 1 if people are willing to help their neighbours Dummy = 1 if neighbours can be counted upon to watch out for kids Dummy = 1 if child has changed residence

14.5

14.5

14.5

10.6

10.7

10.7

1.2% 1.9% 94.8%

1.2% 1.8% 95.0%

1.2% 1.8% 94.9%

96.9%

97.0%

97.0%

5.9%

5.8%

6.0%

41 .7%

44.7%

48.0%

90.3%

90.5%

90.8%

87.9%

88.4%

89.0%

47.9%

55.8%

61 .7%

We study nine components of children's well-being: subjective health; happiness; experience of fear and/or anxiety; relationships with peers, teachers, parents, and siblings; experience of injury; and repetition of a grade. In all cases, the responses are given by the PMK rather than the child. Estimated models are generally ordered probit regressions which allow us to make use of the detailed categorical information about outcome levels available (e.g., a child's health may be reported to be excellent, very good, good, fair, or poor). Probit analysis is used for experience of injury and repetition of a grade because these are dichotomous variables. (Means and frequencies for the dependent variables are reported in Table 8.) Because one cannot directly compare quantitative magnitudes of coefficient estimates across equations when the dependent variables are different,22 and ordered probit coefficients are in any case difficult to interpret directly,23

Social Cohesion and the Well-being of Canadian Children

101

we summarize results for the social relationship variables in Tables 10 through 13.24 Table 9 reports on results that directly include the social support and neighbourhood indices as explanatory variables. For six of ten outcomes, higher social support for the PMK is associated with better outcomes for children. Stronger social support means healthier, happier children who are less anxious and who have better relationships with peers, parents, and teachers. For two outcomes, relationship with siblings and experience of injury, social support is not statistically significant. In one case, repetition of a grade, the coefficient is statistically significant, but with an unexpected sign. Results for the neighbourhood support index are almost identical. The only differences between these two indicators are: (1) relationships with siblings improve with higher levels of neighbourhood support, while this variable was insignificant for parents' social support; and (2) repetition of a grade is not significantly affected by neighbourhood support, while this variable was statistically significant but with an unexpected sign for social support. Table 10 is a sensitivity check. We re-estimate the same models, replacing the social support and neighbourhood indices with dummy variables indicating extremely low levels of either form of support. Results confirm those reported in Table 9. Children whose parents have extremely low levels of social support or neighbourhood support have worse outcomes for almost all indicators studied. Table 11 presents disaggregated results for a few of the particular questions making up the parental support index and neighbourhood support index as well as a question asking about parental involvement in voluntary activities. (To avoid multicollinearity, some questions must be excluded.) The results from specific questions are more spotty, but the parent having someone to trust is an important factor for the child's well-being. Table 11 also presents evidence that there can be a down side to social cohesion. Parental involvement in voluntary activities, generally taken to be an indicator of civic engagement, is positively associated with children's health and school success, but there are negative associations with relationships with parents, siblings, and with the incidence of injury. If volunteering means that Mother or Father go out to do things for and/or with their friends, this may be good for civic spirit and for adult relationships, but from the child's perspective, it may just mean that either Mother or Father is not home. If volunteering means acting as coach, this may mean more time with the child, which could be positive or negative for the relationship between parent and child, depending on the coaching attitude adopted (e.g., a high-pressure 'need to win' attitude on the part of the parent could be very stressful for the child). (The

Table 8 Means/frequencies of 'dependent' variables on social cohesion regressions, 1994 National Longitudinal Survey of Children and Youth Ages 0-11 Child's health

Ages 2-11

1 = excellent 61.1 2 = very good 27.2 3 = good 9.9 4 = fair 1 .5 5 = poor 0.3

Child's happiness

Is child fearful or anxious Gotten along with other kids last 6 months Gotten along with teacher

Gotten along with parents last 6 months

Ages 4-11

1 = happy/ interested in life somewhat happy somewhat unhappy unhappy/ little interest in life life not worth living

89.6 9.1 1.2

1 = very well/ no problems 2 = quite well/ hardly any problems 3 = pretty well /occasional problems 4 = not too well/frequent problems 5 = not well at all/constant problems 1 = very well/ no problems 2 = quite well/ hardly any problems 3 = pretty well /occasional problems 4 = not too well/frequent problems 5 = not well at all/constant problems 1 =very well/ no problems 2 = quite well/ hardly any problems 3 = pretty well /occasional problems

59.1 29.6 10.3 0.9 0.2 77.5 15.9 5.4 1.0 0.2 57.3 30.2 11.6

2= 3= 4= 5=

1 = never/ not true 65.6 2 = sometimes/somewhat true 30.3 3 = often/ very true 4.1

0.1 0.0

Gotten along with siblings last 6 months Has child been injured 1 = yes last 12 months 2 = no Has child repeated a grade

10.3 89.7

4 = not too well/frequent problems 5 = not well at all/constant problems 1 =very well/ no problems 2 = quite well/ hardly any problems 3 = pretty well /occasional problems 4 = not too well/frequent problems 5 = not well at all/constant problems

0.8 0.1 26.7 33.2 33.6 5.8 0.8

1 = yes 2 = no

5.5 94.5

Table 9 The effect of 'social cohesion' indices on current outcomes for children, 1994 National Longitudinal Survey of Children and Youth Gotten along with other kids last 6 months

Gotten along with teacher

Gotten along with parents last 6 months

Gotten along with siblings Iast6 months

Has child been injured last 12 months

Has child repeated a grade

Child's health

Child's happiness

Is child fearful or anxious

Social Support Index

+

+

-

+

+

+

n/s

n/s

+

Neighbourhood Index

+

+

—

+

+

+

+

n/s

n/s

Age of children in sample

0-11 yrs 4-1 1 yrs

2-1 1 yrs

4-11 yrs

4-1 1 yrs

4-11 yrs

4-11 yrs

0-11 yrs

4-11 yrs

Number of observations

19370

15339

12015

10584

12057

10817

19360

8664

Estimation technique

Ordered Ordered probit probit

Ordered probit

Ordered probit

Ordered probit

Ordered probit

Ordered probit

Probit

Probit

12069

Notes: + indicates an improvement in child's well being as 'cohesion' increases; - indicates a reduction in well-being as 'cohesion' increases; n/s indicates no significant relationship at a 10% level. A higher value of either index indicates more support. (See Appendix 1 for details.) The interpretation of the coefficients of an ordered probit model is not straightforward since a positive coefficient will mean an increase in the probability of one or more categories while simultaneously implying a reduction in probability for at least one category.

"able 10 The effect of low levels of 'social cohesion' indices on current outcomes for children, 994 National Longitudinal Survey of Children and Youth

Dummy =1 if low social support Dummy = 1 if low neighbourhood score Age of children in sample Number of observations Estimation technique

Child's health

Child's happiness

Is child fearful or anxious

n/s

-

+

Gotten along with other kids Iast6 months

Gotten along with teacher

Gotten along with parents last 6 months

Gotten along with siblings Iast6 months

Has child been injured last 12 months

Has child repeated a grade

n/s

n/s

+

n/s

-

+

n/s

0-11 yrs 4-11 yrs

2-11 yrs

4-11 yrs

4-11 yrs

4-11 yrs

4-11 yrs

0-11 yrs

4-11 yrs

19370

15339

12015

10584

12057

10817

19360

8664

Ordered probit

Ordered probit

Ordered probit

Ordered probit

Ordered probit

Probit

Probit

12069

Ordered Ordered probit probit

n/s

Notes: + indicates an improvement in child's well being as 'cohesion' increases; - indicates a reduction in well being as 'cohesion' increases; n/s indicates no significant relationship at a 10% level. Low social support means the individual's social support score is less than half the weighted median of the social support index. Low neighbourhood score means the individual's neighbourhood support score is less than half the weighted median of the neighbourhood support index. The interpretation of the coefficients of an ordered probit model is not straightforward since a positive coefficient will mean an increase in the probability of one or more categories while simultaneously implying a reduction in probability for at least one category.

Table 11 The effect of particular 'social cohesion' indicators on current outcomes for children, 1994 National Longitudinal Survey of Children and Youth

Child's health Dummy = 1 if there is someone to trust when having problems Dummy = 1 if there are people to count on in an emergency + Dummy = 1 if no one would help if something went wrong Dummy = 1 if there is any parental involvement in voluntary organizations +

Gotten along with teacher

Gotten along with parents last 6 months

Gotten along with siblings last 6 months

Has child been injured last 12 months

Has child repeated a grade

+

+

+

+

+

n/s

-

-

n/s

n/s

n/s

n/s

n/s

n/s

n/s

-

n/s

n/s

n/s

+

n/s

n/s

n/s

n/s

n/s

-

-

+

-

Child's happiness

Is child fearful or anxious

Gotten along with other kids Iast6 months

n/s

+

+

Dummy = 1 if people are willing to helptheir neighbours + + Dummy = 1 if neighbours can be counted upon to watch out for kids + + Age of children in sample 0-1 1 yrs 4-1 1 yrs 2-1 1 yrs Number of observations 19290 12015 15272 Estimation technique Ordered Ordered Ordered

n/s

n/s

n/s

+

n/s

+

n/s

+

+

+

-

+

4-11 yrs

4-11 yrs

4-1 1 yrs

4-11 yrs

0-11 yrs

4-11 yrs

11961

10538

12003

10765

19280

8626

Ordered

Ordered

Ordered

Ordered

Probit

Probit

Note: + indicates an improvement in child's well being as 'cohesion' increases; - indicates a reduction in well being as 'cohesion' increases; n/s indicates no significant relationship at a 10% level. The interpretation of the coefficients of an ordered probit model is not straightforward since a positive coefficient will mean an increase in the probability of one or more categories while simultaneously implying a reduction in probability for at least one category.

Table 12 The effect of changing residence on current outcomes for children, 1994 National Longitudinal Survey of Children and Youth

Dummy =1 if low social support Dummy = 1 if low neighbourhood score Dummy = 1 if child has changed residence Age of children in sample Number of observations Estimation technique

Child's health

Child's happiness

Is child fearful or anxious

n/s

-

+ +

Gotten along with other kids taste months

Gotten along with teacher

n/s

Gotten along with parents last 6 months

n/s

Gotten along with siblings Iast6 months

-

+

Has child been injured last 12 months

Has child repeated a grade

n/s

n/s

+

n/s

+

+

0-11 yrs 4-11 yrs

2-1 1 yrs 4-1 1 yrs

4-1 1 yrs

4-11 yrs

4-11 yrs

0-11 yrs

4-11 yrs

19370

15339

12015

10584

12057

10817

19360

8664

Ordered probit

Ordered probit

Ordered probit

Ordered probit

Ordered probit

Probit

Probit

12069

Ordered Ordered probit probit

Note: + indicates an improvement in child's well being as 'cohesion' increases; - indicates a reduction in well being as 'cohesion' increases; n/s indicates no significant relationship at a 10% level. Low social support means the individual's social support score is less than half the weighted median of the social support index. Low neighbourhood score means the individual's neighbourhood support score is less than half the weighted median of the neighbourhood support index. The interpretation of the coefficients of an ordered probit model is not straightforward since a positive coefficient will mean an increase in the probability of one or more categories while simultaneously implying a reduction in probability for at least one category.

Social Cohesion and the Well-being of Canadian Children

109

increase in injuries could simply reflect the increase in physical activity with associated risks.) From a policy perspective, an important result of this study is that residential mobility is associated with lower levels of children's well-being. Table 12 adds a dummy variable indicating that the child has changed residence to the specification with the low social support and low neighbourhood support variables. This is the strongest effect observed. In every case, children's outcomes are worse if they have ever changed residence than if they have not. This finding is consistent with the empirical evidence presented by Coleman (1988), Corak (1998), and Hagen et al. (1996). An implication of the findings that children's well-being is consistently and negatively affected by residential moves is the necessity of rethinking the costs and benefits of economic mobility. The literature on migration within Canada has only considered the matter from the perspective of adult incomes,25 but there is also a serious down side to what is generally regarded as an economic good thing (and there may also be losses in well-being for the adults as well as for the children which are not captured here). How large are these effects? Statistical significance does not necessarily mean big changes in outcomes. As mentioned earlier, care must be taken in comparing estimated coefficients across equations, but within one equation we can compare the quantitative impact of one explanatory variable with another. Table 13 uses estimated probit and ordered probit models to calculate the magnitude of social relationship variables relative to poverty or lonemother status on health, peer relationships, repetition of a grade at school, and experience of fear and/or anxiety. We choose these four dimensions of children's well-being because the first three, at least, are aspects of human capital and, hence, fairly standard economic variables (health, education, and social skills). Fear and anxiety is less traditional in economics, but it also seems a compelling choice. For these calculations, other control variables are set at their most common values - thus, a base case child lives with both parents, mother has a post-secondary degree, the child was born when the mother was 25 to 35 years old, the child is male, aged less than 8 years, and has one sibling. We then compute expected changes in child outcomes relative to the base for the following experiments: (1) the family changes residence; (2) the family drops to low levels of social support; (3) the family drops to low levels of neighbourhood support; (4) the family becomes a single-parent family; (5) the family becomes poor.26 Table 13 indicates that the negative impact on children's health associated with changing residence is larger than that associated with single-parent status and of roughly the same magnitude as poverty status. The child's experi-

110 Shelley Phipps Table 13 A comparison of the predicted effects of 'cohesion' variables with effects of 'single parenthood' and 'poverty', 1994 National Longitudinal Survey of Children and Youth

Outcome Question:

Low social support index

Low neighbour hood Single Index mom

Base case*

Moved

Probability

Point Point Point Point Point change change change change change from base from base from base from base from base

In general, would you say your child's health is: excellent During the last 6 months, how well has your child gotten along with other kids: very well, no problems Has your child ever repeated a grade at school? yes How often would you say that your child is too fearful or anxious? never or not true

Poor

0.6498

-0.0384

n/s

-0.1000 -0.0268

-0.037

0.6257

-0.024

-0.115

n/s

-0.0731

n/s

n/s

n/s

0.0070

0.0058

-0.0767

-0.0633 -0.0514

-0.0573

0.0079

0.7023

0.0067

-0.026

*Base case: A household with mom and dad present, which is not poor, the mother has a post secondary degree, the child was born when the mother was between 25 and 35 years old, the child is male and 0-7 years old with one sibling present. The household has not moved nor does it have a low social support index or low neighbourhood index.

ence of fear and/or anxiety is as much affected by lack of social or neighbourhood support as it is by living with a lone parent or living in poverty. The probability of repeating a grade at school is increased as much by changing residence as it is by living with a lone parent or living in poverty. Finally, there are larger negative consequences for peer relationships associated with low parental social support than are associated with loneparent status. To the extent that they match or exceed poverty or single-parent status, the effects associated with social relationships are big and therefore of

Social Cohesion and the Well-being of Canadian Children Figure 7 The quantitative importance of 'social cohesion variables' for child's health now

Note: Nl - Neighbourhood Index Source: National Longitudinal Survey of Children and Youth: Cycle 1, Release 2. Statiscs Canada - Human Resource Development Figure 8 The quantitative importance of 'social cohesion variables' for child's fear and anxiety

Note: SSI - Social Support Index; Nl - Neighbourhood Index Source: National Longitudinal Survey of Children and Youth; Cycle 1, Release 2. Statistics Canada - Human Resource Development Canada

111

112 Shelley Phipps Figure 9 The quantitative importance of 'social cohesion variables' for child's success in school

Source: National Longitudinal Survey of Children and Youth; Cycle 1, Release 2. Statistics Canada - Human Resource Development Canada Figure 10 The quantitative importance of 'social cohesion variables' for child's relationships with other children

Note: SSI - Social Support Index Source: National Longitudinal Survey of Children and Youth; Cycle 1, Release 2. Statistics Canada - Human Resource Development Canada

Social Cohesion and the Well-being of Canadian Children

113

potential policy relevance. (These results are also depicted in bar graph form in Figures 7 through 10.) Conclusions This chapter studies links between social cohesion and the well-being of Canadian children in two main stages. The first part of the chapter is a conceptual essay that argues that both eventual and current outcomes are important. An investment approach, which is the more common in the economics literature, looks for ways in which social relationships will affect the development of children and hence their well-being tomorrow - when they are adults. An alternative, equally important approach looks for links between social relationships and children's well-being now, while they are children. To understand the current well-being of children, a child-centred approach must be adopted; we cannot simply take techniques used to study adults and straightforwardly apply them for children because there are important differences between adults and children. It is argued that children's well-being can be understood as a web of mutually interdependent functionings (e.g., functionings include being healthy, being free of fear and anxiety, having a good education, having good relationships with family and having friends; see Sen 1992). The set of relevant functionings can be expanded as the child develops so that it is equally useful for application to children of different ages, or adults. Very importantly, social relationships can be incorporated in the set of functionings as direct components of well-being (e.g., having good relationships with family or having friends). The chapter points out that children's well-being can also be affected by social relationships among individuals unknown to the child personally. For example, a society in which many adults volunteer to organize cub groups or soccer teams provides important inputs to the current well-being and eventual development prospects of children. (This is in addition to the more often discussed inputs to children's well-being such as income and consumption of goods and services provided privately or publicly.) The second part of the chapter examines the empirical importance of social relationships for the well-being of Canadian children, both while they are children and when they become adults. Using microdata from the 1995 Statistics Canada General Social Survey, the investment perspective asks how past childhood relationships with parents affect the current well-being of adults. Well-being is measured in terms of a set of functionings (income, health, happiness, and quality of relationships with spouse, children, and

114 Shelley Phipps parents). Results indicate that good childhood relationships with parents significantly improve current well-being, controlling for other relevant personal characteristics. Adopting the approach that children count now, the 1994 National Longitudinal Survey of Children and Youth is then used to test the impact of three social cohesion variables on current child functionings. Social support of the parent, social cohesion in the neighbourhood, and residential moves all have statistically significant relationships with children's well-being. (Functionings studied include health, happiness, quality of relationships with parents, siblings, peers, teachers, and repetition of a grade at school.) These empirical findings, while exploratory in nature, suggest that more work in this area is required. Not only are these social cohesion variables found to have statistically significant links with children's well-being, but their influence is as big as those associated with low-income status or single parenthood, for example. The finding that children's well-being is consistently and negatively affected by residential moves illustrates that a consideration of social cohesion can affect standard economic policy conclusions: migration may not be as unambiguously a good thing as the economics literature has to date suggested. Appendix Calculation of Social Support Index Score Questions used in the calculation: DO YOU STRONGLY DISAGREE, DISAGREE, AGREE, OR STRONGLY AGREE WITH THE FOLLOWING STATEMENT: 1 If something went wrong, no one would help me. 2 I have family and friends who help me feel safe, secure, and happy. 3 There is someone I trust whom I would turn to for advice if I were having problems. 4 There is no one I feel comfortable talking about problems with. 5 I lack a feeling of closeness with another person. 6 There are people I can count on in an emergency. Item values: Strongly disagree Disagree Agree Strongly agree

(0) (1) (2) (3)

Social Cohesion and the Well-being of Canadian Children

115

Item values were reversed for questions 1, 4, and 5. The factor score was calculated using these items unweighted. The total score varies between 0 and 18, a high score indicating the presence of social support. Calculation of Neighbourhood Index Score Questions used in the calculation: DO YOU STRONGLY AGREE, AGREE, DISAGREE, OR STRONGLY DISAGREE WITH THE FOLLOWING STATEMENT: 1 2 3 4

If there is a problem around here, the neighbours get together to deal with it. There are adults in the neighbourhood that children can look up to. People around here are willing to help their neighbours. You can count on adults in the neighbourhood to watch out that children are safe and don't get in trouble. 5 When I'm away from home, I know that my neighbours will keep their eyes open for possible trouble. Item values: Strongly disagree Disagree Agree Strongly agree

(0) (1) (2) (3)

The factor score was calculated using these items weighted. The total score varies between 0 and 15, a high score indicating a high degree of neighbour cohesiveness. Notes 1 I would like to thank Lynn Lethbridge, Michael Rushe, and Kara Beckles for excellent research assistance, Peter Burton, Jeff Dayton-Johnson, and Lars Osberg for extremely helpful comments, and the Department of Canadian Heritage for financial support. 2 This literature will be discussed in more detail below. 3 There is a very large literature that studies child poverty. See, e.g., Picot et al. (1999); Smeeding and Sullivan (1998). 4 Rawls himself included liberty and opportunity, income and wealth as the basis of self-respect (1971: 303).

116 Shelley Phipps 5 This approach is very similar to that advocated by Allardt (1981), who talks about 'having, loving and being' as the key components of well-being. The work of Klasen (1998) and Ross et al. (1996) are examples of empirical studies that essentially adopt a functionings approach. 6 This is in contrast, e.g., to research focusing on the consequences of social cohesion for macroeconomic growth (Helliwell and Putnam 1995; Knack and Keefer 1997) or on employment (e.g., Britton 1998). 7 These researchers are said to study the ecology of human development. 8 There is already a large empirical literature, much of it outside economics, demonstrating that social relationships affect outcomes for children. See e.g., Amato et al. (1995); Amato (1994), Cherlin et al. (1995), Furstenberg and Teitler (1994), Gauze et al. (1996), Hagan et al. (1996), Haveman et al. (1991), Parcel and Menaghan (1993; 1994), Upperman and Gauthier (1998). 9 It is also important to note the possibility of negative consequences associated with, e.g., stifling norms from a very cohesive community. 10 Norms may be positive or negative. 11 Behrman et al. (1995), e.g., adopt a similar perspective. 12 This is not to say that the functionings approach is not also well suited for the study of adult well-being and for many of the same reasons. 13 As mentioned earlier, although the general tone of this chapter suggests that greater social cohesion will be beneficial for children, this need not always be true. Group norms, e.g., may be positive or negative with respect to child-rearing practices. 14 The choice of the six functionings represented in Figure 2 is not central to the argument. However, these functionings do have the advantage of connecting to the empirical work to follow in the second part of this chapter. 15 In the GSS, household income is reported in twelve categories, with zero income being the bottom category and greater than $100,000 being the top category. We transform from categories to a continuous variable, by assigning each individual the mid-point of the income interval in which his or her income falls. For those with incomes greater than $100,000, we assign the average income for individuals with greater than $100,000, as calculated using Survey of Consumer Finance data for the same year ($155,000). 16 Notice that if it is true that individuals who are happier because they have children are better parents or have better relationships with their children, then we are investigating the possibility of multigenerational effects here. 17 This is not the most desirable measure. It could simply reflect geographic proximity or the need and/or dependency of the parent. 18 Full regression results are available in a longer version of this chapter which is available upon request from the author.

Social Cohesion and the Well-being of Canadian Children

117

19 One potential problem with all of the econometric results reported in this section is that because current circumstances may cause individuals to look back with either overly fond or overly bitter memories, it is possible that individuals for whom things are currently not going very well in life blame their parents and, hence, report poorer relationships or less happiness during childhood than they otherwise might (and vice versa). An appropriate solution to this problem, which is outside the scope of this project, would be to instrument the relationship with Mother and Father variables, using explanatory variables correlated with the relationship variables, but not with the current outcome indicators. 20 Our choice of specification for the basic control variables follows the excellent work of Dooley et al. (1998). 21 For example: (1) 'If something went wrong, no one would help me.' (2) 'There is someone I trust whom I would turn to for advice if I were having problems.' (3) There are people I can count on in an emergency.' 22 One cannot equate, e.g., a ' 1' meaning 'excellent health' with a ' 1' meaning the child gets along with siblings 'very well/with no problems.' 23 A positive coefficient in an ordered probit model means that the distribution of answers shifts to the right. Hence, we know that the probability of being in the lowest category (e.g., of health) decreases and the probability of being in the highest category increases. It is not possible to discern the changes in the distribution of the middle categories from eyeballing the coefficients. 24 Detailed parameter estimates are available upon request from the author. Results obtained for the basic control variables are consistent with previous work (e.g., Dooley et al. 1998). 25 See, e.g., Day (1992), Robinson and Tomes (1982), Osberg et al. (1994). 26 This thought experiment does not simulate the dynamics associated with any of these changes - we simply compare expected outcome probabilities for children in the two situations.

References Allardt, Erik 1981. 'Experiences from the Comparative Scandinavian Welfare Study, with a Bibliography of the Project.' European Journal of Political Research 9: 101-11. - 1993. 'Having, Loving, Being: An Alternative to the Swedish Model of Welfare Research.' In Martha Nussbaum and Sen Amartya, eds., The Quality of Life. World Institute for Development Economics. Oxford: Oxford University Press, 88-94. Amato, Paul R. 1994. 'Life-Span Adjustment of Children to Their Parents' Divorce.' Future of Children 4(1): 143-64. Amato, Paul R., Laura Spencer Loomis, and Alan Booth. 1995. 'Parental Divorce,

118 Shelley Phipps Marital Conflict, and Offspring Well-Being During Early Adulthood.' Social Forces 73(3): 895-915. Becker, Gary S. 1991. A Treatise on the Family. Cambridge, Mass.: Harvard University Press. Becker, Gary S., and Nigel Tomes. 1979. 'An Equilibrium Theory of the Distribution of Income and Intergenerational Mobility.' Journal of Political Economy 87(6): 1153-89. - 1986. 'Human Capital and the Rise and Fall of Families.' Journal of Labour Economics 4(3): S1-S39. Behrman, Jere R., Robert A. Pollack, and Paul Taubman. 1995. From Parent to Child: Intrahousehold Allocations and Intergenerational Relations in the United States. Chicago: University of Chicago Press. Britton, Andrew. 1998. 'Employment and Social Cohesion.' In A.B. Atkinson and John Hills, eds., Exclusion, Employment and Opportunity, Centre for Analysis of Social Exclusion, London School of Economics 4: 21-30. Bronfenbrenner, Urie. 1979. The Ecology of Human Development: Experiments by Nature and Design. Cambridge, Mass.: Harvard University Press. Browning, Martin. 1992. 'Children and Household Economic Behavior.' Journal of Economic Literature 30(Sept.): 1434-75. Cherlin, Andrew, Kathleen E. Kieraan, and P.L. Lindsay Chase-Lansdale. 1995. 'Parental Divorce in Childhood and Demograhic Outcomes in Young Adulthood.' Demography 32(3): 299-318. Cochran, Moncrieff, Mary Larner, David Riley, Lars Gunnarsson, and Charles R. Henderson Jr. 1990. Extending Families: The Social Networks of Parents and Their Children. New York: Cambridge University Press. Coleman, James S. 1988. 'Social Human Capital in the Creation of Human Capital.' American Journal of Sociology 94(Suppl.): S95-S120. Corak, Miles. 1998. 'Getting Ahead in Life: Does Your Parents' Income Count?' Canadian Social Trends. Ottawa: Statistics Canada, Cat. 11-008-XPE, 6-10. Curtis, Lori, Martin Dooley, and Shelley Phipps. 2000. 'Does Either Mother or Father Know Best? An Assessment of Parent/Child Agreement in the NLSCY.' Mimeo. Day, Kathleen. 1992. 'Interprovincial Migration and Local Public Goods.' Canadian Journal of Economics 25(1): 123-44. Deaton, Angus S., and John Muellbauer. 1986. 'On Measuring Child Costs: With Applications to Poor Countries.' 94(4): 720-44. Dooley, Martin D., Lori Curtis, Ellen L. Lipman, and David H. Feeny. 1998. 'Child Psychiatric Disorders, Poor School Performance and Social Problems: The Roles of Family Structure and Low Income in Cycle One of the National Longitudinal Survey of Children and Youth.' In Miles Corak, ed., Labour markets, Social

Social Cohesion and the Well-being of Canadian Children

119

Institutions, and the Future of Canada's Children. Chapter 7. Ottawa: Statistics Canada, Cat. 89-553-XPB. Furstenberg, Frank F. Jr., and Julien O. Teitler. 1994. 'Reconsidering the Effects of Marital Disruption: What Happens to Children of Divorce in Early Adulthood?' Journal of Family Issues 15(2): 173-90. Garbarino, James, 1979. 'The Issue is Human Quality: In Praise of Children.' Children and Youth Services Review 1: 353-77. Gauze, Cyma, William M. Bukowski, Jasmin Aquan-Assee, and Lorrie K. Sippola. 1996. 'Interactions between Family Environment and Friendship and Associations with Self-Perceived Well-Being during Early Adolescence.' Child Development, 67: 2201-16. Hagen, John, Ross MacMillan, and Blair Wheaton. 1996. 'New Kid in Town: Social Capital and the Life Course Effects of Family Migration on Children.' American Sociological Review 61 (June): 368-85. Hagenaars, A.J.M. 1986. The Perception of Poverty. Amersterdam: North-Holland. Haveman, Robert, and Barbara Wolfe. 1995. 'The Determinants of Children's Attainments: A Review of Methods and Findings.' Journal of Economic Literature 33: 1829-78. Haveman, R., Barbara Wolfe, and James Spaulding. 1991. 'Childhood Events and Circumstances Influencing High School Completion.' Demography 28(1): 133-57. Helliwell, John F., and Robert D. Putnam. 1995. 'Economic Growth and Social Capital in Italy.' Eastern Economic Journal 21(3): 295-307. Klasen, Stephan. 1998. 'Measuring Poverty and Deprivation in South Africa.' Unpublished paper. Knack, Stephen, and Philip Keefer. 1997. 'Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.' Quarterly Journal of Economics 112(4): 1251-88. Leibowitz, Arleen. 1974. 'Home Investments in Children.' Journal of Political Economy 82(2): S111-S131. Nakamura, Alice, and Nakamura, Masao. 1992. The Econometrics of Female Labor Supply and Children.' Econometric Reviews 11(1): 1-71. Osberg, Lars, Daniel Gordon, and Zhengxi Lin. 1994. 'Interregional Migration and Interindustry Labour Mobility in Canada: A Simultaneous Approach.' Canadian Journal of Economics 27(1): 58-79. Parcel, Toby L., and Elizabeth G. Menaghan. 1993. 'Family Social Capital and Children's Behaviour Problems.' Social Psychology Quarterly 56(2): 120-35. - 1994. 'Early Parental Work, Family Social Capital, and Early Childhood Outcomes.' American Journal of Sociology 99(4): 972-1009. Phipps, Shelley A. 1999. 'Economics and the Well-being of Canadian Children.' Canadian Journal of Economics 32(5): 1135-63.

120

Shelley Phipps

Phipps, Shelley A., and Peter Burton. 1995. 'Sharing within Families: Implications for the Measurement of Poverty Among Individuals in Canada.' Canadian Journal of Economics 28(1): 177-204. - 1996. 'Collective Models of Family Behavior: Implications for Economic Policy.' Canadian Public Policy 22(2): 129^3. Picot, G.M. Zyblock, and W. Pyper. 1999. 'Why Do Children Move into and out of Low Income? Changing Labour Market Conditions or Marriage and Divorce?' Ottawa: Statistics Canada Analytical Studies Branch Research Paper 132. Putnam, Robert D. 1995. 'Bowling Alone, Revisited.' Responsive Community (Spring): 18-33. Qvortrup, Jens. 1990. Childhood as a Social Phenomenon - An Introduction to a Series of National Reports. New York: Delacourt Press. Rawls, John. 1971. A Theory of Justice. Cambridge, Mass.: Harvard University Press. Robinson, Chris, and Nigel Tomes. 1982. 'Self-Selection and Interprovincial Migration in Canada.' Canadian Journal of Economics 15(3): 474-502. Ross, David P., Katherine Scott, and Mark Kelly. 1996. Child Poverty: What are the Consequences? Ottawa: Canadian Council on Social Development. Sen, Amartya. 1992. Inequality Re-examined. Cambridge, Mass.: Harvard University Press. Smeeding, Timothy M., and Dennis H. Sullivan. 1998. 'Generations and the Distribution of Economic Weil-Being: A Cross-National View.' American Economic Review 88(2): 254-8. Stanley, Dick. 1997. 'The Economic Consequences of Social Cohesion.' Hull, Que.: Department of Heritage Canada. Unpublished paper. Upperman, Kate, and Anne Helene Gauthier. 1998. 'What Makes a Difference for Children? Social Capital and Neighbourhood Characteristics.' Policy Options (Sept.): 24-7.

Social Cohesion and Health JOHN N. LAVIS AND GREGORY L. STODDART1

Until recently, the topic of social cohesion and the topic of health were rarely mentioned together. Now, the two constructs seem inseparable, with social cohesion typically posited to be a cause and health an effect. More social cohesion has been posited to lead to 'more' health; less social cohesion has been posited to lead to 'less' health. But whether this is actually true, and if so, what explains this relationship can be difficult to discern. It may be that health is the cause. Or, it may be that social cohesion and health are being influenced by a third factor. A jurisdiction with a very unequal distribution of income, for example, may have low levels of both social cohesion and health, and thus represent a case of income distribution determining whether and in what way social cohesion and health levels change. This chapter examines the relationship between social cohesion and health.2 In the following section we review terms and provide basic descriptive data. In the second section, we explore social cohesion and health as a relationship and present the results of new empirical research regarding this relationship in the group of seven (G7) countries (Canada, the United States, the United Kingdom, France, Germany, Italy, and Japan). For the empirical analysis we used the World Values Surveys (WVS) cross-national time-series data (see Inglehart 1996; Nevitte 1996; Helliwell 1996). The first WVS was conducted in 1981, the other in 1990.3 We restricted our analyses to the G7 to provide a reference group of countries at a similar level of economic development as Canada. In the third section of this chapter we synthesize and critically appraise empirical research to inform our discussions about the strength of (some of) these relationships, specifically those involving selected pathways through the determinants of health, namely, income inequalities, employment and working conditions, and social support. In the final section we draw three main conclusions from our endeavour.

122 John N. Lavis and Gregory L. Stoddart Background We begin by reviewing terms and providing basic descriptive data. For social cohesion and for health, we review formal definitions, lay interpretations (where relevant), frequently used indicators, and (when available) a description of their current levels using the World Values Surveys. Indicators should follow from formal definitions. Many definitions of social cohesion, however, incorporate normative statements about the appropriate (or inappropriate) level of one or more indicators that may be associated with social cohesion, while not capturing the construct directly (e.g., income inequality). We avoid this circularity by distinguishing between definition and indicators. We also recognize that the interrelationships between formal definitions, lay interpretations, and indicators can serve important purposes in policy making (Stone 1997). Social Cohesion Social cohesion provides a challenging starting point for a discussion of terms. As noted, formal definitions often mix definition with indicator: for example, 'social cohesion involves ... reducing disparities in wealth and income' (Maxwell 1996). Recognizing the difficulties inherent in defining a contested concept like social cohesion, Jenson (1998) created a typology of five dimensions of paired concepts related to social cohesion: belonging/ isolation, inclusion/exclusion, participation/non-involvement, recognition/rejection, and legitimacy/illegitimacy. This typology brings to mind the anecdote about the blind person who arrives at a different conclusion about what an elephant is depending on what part of the elephant the person touches. We define social cohesion as the networks, norms and trust that bring people together to take action. Social cohesion is the glue that binds people together. The network dimension of social cohesion embodies both horizontal associations between people (uniting ties within groups) and vertical associations across social divides (bridging ties across groups). As the World Bank's web site for social capital4 states, horizontal ties are needed to give groups a sense of identity and common purpose; however, without vertical or bridging ties, horizontal ties can become a basis for the pursuit of narrow interests. The mafia illustrates this point well. As well, the norms and trust dimension of social cohesion embodies transferable, not situation-specific, norms and trust. Despite this breadth of concepts falling under the social cohesion banner, the indicators most often used are remarkably narrow. In addition, indicators

Social Cohesion and Health

123

Table 1 Descriptive statistics for Canada in 1981 and 1990

Measures Measures of social cohesion Proportion who trust others Mean number of memberships Proportion in religious or church organizations Proportion in sports and cultural organizations Proportion in organizations concerned with health3 Proportion in political groups Proportion in a trade union Measure of health Proportion in good health

1981 (N1=1254)

1990 (N2=1730)

0.50 1.03 0.33 0.25 0.21 0.11

0.53 1.68* 0.25* 0.48* 0.09 0.32* 0.12

0.81

0.80

na

Source: World Values Survey, as extracted from chass.utoronto.ca (file location: data/ icp117; file name: world.va193.da; file date: 25 April 1995). a

Organizations concerned with health are a sub-group of sports and cultural organizations, however, we calculated the proportion of respondents who are members of this sub-group using the entire sample as the denominator. * Significant at the 95% level using a two-sample f-test for differences between 1981 and 1990.

measure social cohesion at the individual, not the aggregate, level. The two most commonly used indicators are trust (typically measured using a single question such as, 'Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?') and number of memberships in voluntary organizations. Use of both of these indicators can be traced back to a cross-national comparison of social cohesion made in the early 1960s by Almond and Verba (1963). From the World Values Surveys we were able to ascertain three things: (1) how the level of social cohesion has changed over time in Canada; (2) what individual-level factors5 have been associated with high levels of social cohesion in Canada; and (3) how the level of social cohesion in Canada compares with the levels in other G7 countries. Between 1981 and 1990, the proportion of respondents in Canada who trust others remained about the same, and the mean number of association memberships increased significantly (Table 1). More specifically, membership in sports and cultural organizations and in political groups increased significantly, while membership in religious or church organizations decreased significantly (with the former more than offsetting the latter). Following previous work, we did not analyse these results

124 John N. Lavis and Gregory L. Stoddart separately according to the reason given for membership, although we recognize that some reasons (e.g., 'to make a contribution to my local community') better reflect our definition of social cohesion than other reasons (e.g., 'I did not want to, but I could not refuse'). Older, white, high-income, educated, and employed respondents tended to have the highest levels of trust and to have a larger number of memberships in voluntary organizations (Table 2). Racial differences were particularly pronounced for both measures of social cohesion, while sex differences were not. As has been found in studies dating back to the early 1960s (e.g., Almond and Verba 1963), levels of trust were higher in Canada, the United States, and the United Kingdom than in France, Germany, and Italy, with Japan intermediate between these two groupings (Table 3). The number of memberships followed a slightly different pattern, with Germany in 1990 looking more like Canada, the United States, and the United Kingdom, and with Japan having the lowest mean number of memberships. Health The challenge of establishing meaningful terms does not end with social cohesion; defining health also presents a challenge. Formal definitions range from biological (e.g., neurotransmitter levels) to clinical (e.g., absence of disease) to holistic (e.g., well-being). Economist Robert Evans has suggested (rather dryly) that, from a physician's perspective, health is a state of incomplete diagnosis. From a policy perspective, pragmatism often drives the choice of a definition of health as 'absence of disease,' while idealism often drives the choice of the World Health Organization's definition of health as 'the ability to identify and to realize aspirations, to satisfy needs, and to change or cope with the environment. Health is therefore a resource for everyday life, not the objective of living. Health is a positive concept emphasizing social and personal resources, as well as physical capacities' (World Health Organization 1986). Lay interpretations of health also span a range: from 'not being ill, sick or disabled' to 'feeling good, energetic and strong' to 'being able to do what one likes' (Noack 1997). Rather than choose one definition over another, we follow Evans and Stoddart (1990) in adopting an organizing framework for health indicators and consider their full breadth. This framework includes an individual's response to health determinants and the outcomes (disease, health and function,6 and well-being) that follow from these responses. An individual's response can include both behavioural and biological elements, and the response may be health enhancing or health damaging, inherited or acquired. Behavioural

Social Cohesion and Health 125 Table 2 Cross-tabulations for Canada in 1981 (weighted) and 1990 (weighted) Measures of social cohesion and health Number of memberships

Trust Measures

Good health

1981

1990

1981

1990

1981

1990

Age (N,=1236) (N2=1704)

45+ 18-44

0.51 0.49

0.54 0.51

1.12 0.98

1.77 1.61

0.72 0.88

0.68 0.88

Sex (N,=1251) (N2=1730)

0.49 Male Female 0.50

0.52 0.53

1.03 1.04

1.64 1.72

0.80 0.82

0.81 0.79

Race (N1=779) (N2=1608)

Black White

0.32 0.61

0.36 0.53

0.63 1.21

0.86 1.71

0.77 0.84

0.95 0.80

Married (N1=1254) (N2=1728)

No Yes

0.46 0.51

0.47 0.55

0.90 1.09

1.57 1.74

0.81 0.81

0.78 0.81

Income (N,=1029) (N2=1461)

$30K

0.48 0.54

0.42 0.56

0.93 1.27

1.12 1.87

0.78 0.89

0.62 0.85

Education (N1=1227) (N2=1690)

12yrs

0.40 0.54

0.38 0.57

0.76 1.18

1.10 1.84

0.69 0.86

0.64 0.85

Emp grade (N,=1222) (N2=1629)

Manual 0.46 Nonman 0.56

0.45 0.60

0.92 1.25

1.29 2.05

0.77 0.88

0.75 0.85

Employment (N1=795) (N2=1179)

no yes

0.37 0.53

0.44 0.54

0.64 1.16

0.95 1.87

0.83 0.87

0.78 0.88

Source: World Values Survey, as extracted from chass.utoronto.ca (file location: data/ icp117; file name: world.va193.da; file date: 25 April 1995). Note: The small number of 'black' respondents in the 1990 survey may not be representative of all potential 'black' respondents. This may explain the finding that more 'black' respondents reported good health in that year than 'white' respondents, a result at odds with the finding from 1981 and from other studies.

126 John N. Lavis and Gregory L. Stoddart Table 3

Descriptive statistics for G7 countries in 1981 (weighted) and 1990 (weighted) Measures of social cohesion and health

Trust

Number of memberships

Good health

Country

1981

1990

1981

1990

1981

1990

Canada (N1=1254) (N2=1730)

0.50

0.53

1.03

1.68

0.81

0.80

United States (N1=2325) (N2=1839)

0.45

0.50

1.50

1.94

0.77

0.79

United Kingdom (N1=1231) (N2=1484)

0.44

0.44

0.92

1.18

0.72

0.75

France (N1=1200) (N2=1002)

0.25

0.23

0.42

0.73

0.59

0.66

Germany (N1=1305) (N2=2101)

0.30

0.38

0.74

1.37

0.52

0.56

Italy (N1=1348) (N2=2010)

0.26

0.37

0.39

0.76

0.51

0.60

Japan (N,=1204) (N2=1011)

0.41

0.42

0.37

0.49

0.44

0.44

Source: World Values Survey, as extracted from chass.utoronto.ca (file location: data/ icpl 17; file name: world.val93.da; file date: 25 April 1995).

responses cover a range of meaningful categories of action, from consumption (e.g., smoking, drinking, drug use) to help seeking (e.g., seeking social support, visiting a health care provider, contacting a local politician) and avoidance (e.g., repeatedly missing days of work to avoid work-related stressors). These behaviours reflect, in part, the coping strategies that individuals use to respond to their social environment, and the choice of behaviour and extent to which it is engaged in reflect the personal and financial resources that individuals have amassed in the context of their environments and en-

Social Cohesion and Health 127 dowments. Biological responses include, for example, neuroendocrine, immunologic, and physiologic responses (Evans et al. 1994). In the Evans and Stoddart (1990) framework, outcomes include disease and disability, health and function, and a broader but related outcome: wellbeing. The first category of health outcome, disease and disability, includes professionally designated medical conditions. The second category, mental and physical health and function, represents an individual's rather than a health care provider's description of his or her health and the value or importance to the individual of that health. Both of these concepts are, in turn, distinguished from a higher-order concept, well-being, which represents an individual's interpretation of his or her general happiness. Health is assumed to be one of the constituent elements of well-being, but individuals may be willing to sacrifice some gains in health status for more of something else, like a higher income, which may translate more directly into improved wellbeing. Self-reported health status, a health indicator used in the research that we discuss in this chapter, represents an example of mental and physical health and function combined. This indicator is based on an individual's rating of his or her own health on a five-point Likert scale, typically including poor, fair, good, very good, and excellent. This simple indicator has good predictive capacity, having been shown in prospective studies to be correlated with future health care use (Fylkesnes 1993; Hulka and Wheat 1985; Krakau 1991) and with subsequent mortality (see Idler and Benyamini 1997 for a review of twenty-seven studies). Self-ratings have been found to reflect serious, chronic conditions, but self-ratings are not affected by acute, transitory illnesses (Goldstein et al. 1984; Pope 1988). As well, this measure has been shown to be correlated with more complex health indices (Rowan 1994) and to have good test-retest reliability (Lundberg and Manderbacka 1996; Martikainen et al. 1996). From the World Values Surveys we were able to ascertain how the level of self-reported health status has changed over time in Canada, what individuallevel factors have been associated with high levels of health status in Canada, and how the level of health status in Canada compares with the levels in other G7 countries. Between 1981 and 1990, the proportions of respondents in Canada who reported their health as very good or good (the top two categories of the five provided in the surveys) remained about the same (Table 1). Younger, white, high-income, educated, high employment grade, and employed respondents were more likely to report good health (Table 2). Levels of good health in G7 countries generally followed a similar pattern to levels of trust, with levels of good health higher in Canada, the United States, and

128 John N. Lavis and Gregory L. Stoddart the United Kingdom than in France, Germany, and Italy (Table 3). Japan provides an exception to this general association between levels of trust and levels of good health: Japan had an intermediate level of trust, but a relatively low health status (at least using this health indicator).7 Social Cohesion and Health Social cohesion and health may be causally related. Quite plausibly, for example, more socially cohesive communities (however communities are defined) could provide visits to frail elderly people who may become ill or injured if left alone. They could provide shelters for homeless people who may suffer from hypothermia during the winter. Socially cohesive communities may have lower rates of violent injuries (like gunshot wounds) than less socially cohesive communities. Several studies have examined the association between social cohesion and health. They have used various approaches, ranging from case studies to quantitative studies that can statistically adjust for potential risk factors for ill health that are unrelated to social cohesion. All have found significant associations between social cohesion and health (Kawachi 1998), although none have addressed the mechanisms through which these relationships might operate. The most frequently cited case study involves the town of Roseto, Pennsylvania. In the 1960s, close family ties and cohesive community relationships in Roseto appeared to offer a protective effect against death from heart disease: a neighbouring town with, for example, less than half the number of civic organizations per capita as Roseto, had roughly double the rate of death from heart disease, despite similar smoking rates, overweight and sedentary profiles, and diets (Bruhn and Wolf 1979; Wolf and Bruhn 1992). Moreover, as the density of civic organizations and other measures of social cohesion declined in Roseto over three decades, the town's rate of death from heart disease began to look more like the rate in neighbouring towns (Bruhn and Wolf 1979; Egolf et al. 1992). To address the possibility that factors other than social cohesion could explain the 'Roseto effect,' Kawachi et al. (1997) and Sampson et al. (1997) quantitatively assessed the association between measures of social cohesion and health-related outcomes and adjusted for potential individual-level risk factors for ill health. The Kawachi group examined three measures of social cohesion - trust, membership in voluntary organizations, and norms of reciprocity - each ascertained first the individual level through surveys and then aggregated to the state level. All three measures correlated strongly with death rates. For example, a one-unit increment in the average per capita group membership at the state level was associated with a lower, age-adjusted

Social Cohesion and Health

129

overall death rate of 66.8 deaths per 100,000 state residents, after adjusting for household poverty rates. That is, assuming that the relationship between membership in voluntary organizations and health is causal, if each state resident joined one new voluntary organization, about 67 fewer people would die for every 100,000 state residents (Kawachi et al. 1997). Sampson et al. (1997) produced a summary index of collective efficacy, developed by combining responses to questions about trust and other measures of social cohesion, each asked first at the individual level through surveys and then with the responses aggregated to the neighbourhood level. The summary index turned out to be correlated with average levels of membership in voluntary organizations at the neighbourhood level. Collective efficacy was significantly associated with perceived neighbourhood violence and with violent victimization, even after adjusting for individual-level measures of age, sex, ethnicity, marital status, socioeconomic status, home ownership, and years of residence in the neighbourhood. A similar pattern was seen for homicide rates: an elevation of two standard deviations in neighbourhood collective efficacy was associated with a 40 per cent reduction in the expected homicide rate. This innovative use of data from both the individual and aggregate levels to address competing explanations was also pursued in a study using selfreported health status as the outcome (Kawachi et al. unpublished). The investigators examined three measures of social cohesion - trust, membership in voluntary organizations, and norms of reciprocity - again each ascertained first at the individual level through surveys and then aggregated to the state level. Each measure was strongly correlated with individual self-reported health status, even after adjusting for individual-level risk factors for ill health, like low household income, limited educational attainment, living alone, lack of health insurance coverage, smoking, and obesity. For example, the odds of being in fair or poor health (compared with excellent or very good health) if living in states with the lowest levels of trust was 1.41 times higher than if living in states with the highest levels of trust, adjusting for individual-level risk factors. Recognizing the limitations of its cross-sectional design and individuallevel measures of social cohesion, we used the World Values Surveys and multivariate logistic regressions to examine one aspect of these relationships - the association between trust or membership(s) in voluntary organizations and self-reported health status. In keeping with our expectations, trust was positively associated with good health in Canada, the United States, Germany, Italy, and Japan, even after controlling for potential individual-level risk factors for which we had data (Table 4 shows the odds-ratio results for

130 John N. Lavis and Gregory L. Stoddart Table 4 Logistic regressions for G7 countries in 1990 (weighted) Country (and independent variables) Canada Trust Trust plus No. memb No. memb plus United States Trust Trust plus No. memb No. memb plus United Kingdom Trust Trust plus No. memb No. memb plus France Trust Trust plus No. memb No. memb plus Germany Trust Trust plus No. memb No. memb plus Italy Trust Trust plus No. memb No. memb plus Japan Trust Trust plus No. memb No. memb plus

Odds ratios for good health

(N2=1667) (N2=1333) (N2=1723) (N2=1372)

1.73(1.35-2.22) 1.54(1.15-2.08)" 1.05(0.97-1.14) 1.01 (0.93-1.10)"

(N2=1781) (N2=1542) (N2=1838) (N2=1588)

1.57(1.24-1.99) 1.64(1. 25-2.1 6) 1.08(1.02-1.15) 1. 03 (0.96-1. 11 )•

(N2=1439) (N2=1425) (N2=1482) (N2=1446)

1.24(0.95-1.63) 1.31 (0.98-1.75)" 1.12(1.01-1.23) 1.07(0.97-1.18)°

(N2=937) (N2=937) (N2=1000) (N2=921)

1.73(1.23-2.45) 1.57(1.10-2.24)" 1.08(0.97-1.21) 1.08 (0.96-1 .22)d

(N2=1723) (N2=1582) (N2=2097) (N2=1928)

1.92(1.66-2.35) 1.82(1.43-2.30)a 1.07(1.00-1.14) 1.03 (0.96-1. 10)a

(N2=1914) (N2=1896) (N2=1997) (N2=1979)

1.72(1.34-2.21) 1.53(1.18-1.99)° 1.10(0.99-1.23) 0.99(0.89-1.10)°

(N2=903) (N2=895) (N2=997) (N2=989)

1.62(1.24-2.11) 1.53(1.16-2.00)° 1.13(0.99-1.29) 1.12(0.97-1.28)°

Source: World Values Survey, as extracted from chass.utoronto.ca (file location: data/ icp117; file name: world.val 93.da; file date: 25 April 1995). a Controlling for age, income, education, and employment grade. b Controlling for age and education. c Controlling for age and employment grade. d Controlling for age, education, and employment grade.

Social Cohesion and Health

131

good health). The relationships were in the predicted direction, but of borderline statistical significance, in the United Kingdom and France. Thus, we conclude that trust is highly correlated with good health, although we cannot say that trust causes good health. As well, if this relationship is causal, we cannot determine the pathways through which it operates. The number of memberships was not associated with good health. Some Specific Pathways through Health Determinants Rather than do an injustice to the many possible pathways through which social cohesion may affect health, we focus the remainder of our review of empirical research on selected pathways involving the determinants of health. With this focus we can do some justice to determining the direction and magnitude of the relationships that involve social cohesion and health, thereby moving, if only figuratively, towards a measure of the social cohesion elasticity of health. By health determinants we mean the factors that determine why some people are healthy and others not (Evans and Stoddart 1990; Evans et al. 1994). These factors can include social environments (e.g., early childhood development, income distribution, employment and working conditions, social supports, or cultural diversity), physical environments (e.g., pollution, public spaces, safety and crime, or homelessness), and genetic endowments (e.g., physiological capacity to buffer stress responses), as well as health care. Alternatively, following the Federal/Provincial/Territorial Advisory Committee on Population Health (1994), these factors can be grouped into nine categories: healthy child development, education, income and social status, employment and working conditions, social support network, physical environment, biological and genetic endowment, personal health practices and coping skills, and health services. An ever-growing literature has demonstrated powerful relationships between such 'upstream' factors and health (Evans et al. 1994; Amick et al. 1995; Blane et al. 1996). In this chapter, we focus primarily on income distribution, employment and working conditions, and social support. These factors have been rigorously studied and can be addressed by government action (Osberg 1998). Large income inequalities, adverse labour market experiences, and limited social supports have repeatedly been shown to affect health (Lynch et al. 1997; Lavis 1998; Berkman 1995). Moreover, governments can affect the degree of income inequality within a country, the nature and distribution of labour market experiences like unemployment, job insecurity, and job strain, and the nature and distribution of social supports. Again, space limitations

132 John N. Lavis and Gregory L. Stoddart mean that even by narrowing our review to pathways involving these selected health determinants, we cannot be exhaustive in our treatment of them. The full story is even more complex than what we can reasonably address in this chapter and includes, among other things, complex interactions among categories of health determinants and dynamic issues such as latency effects (Evans et al. 1994; Hertzman 1994; Frank 1995; Stoddart 1995). Income Inequalities Many industrialized countries have experienced growing income inequalities over the past decade, and these income inequalities can have profound health consequences. The literature on the links between income inequalities and health provides an illustrative example of the influence of the social environment on health. Like all research, many of the studies have methodological limitations, most significantly income-related biases in response rates to income surveys (Judge 1995; with a response by Wilkinson 1995) and potential for the results to be attributable to differences at baseline (e.g., in health) between those at the lower end of the income distribution compared with those at the top. Taken together, however, the full range of studies is quite persuasive. Studies have examined a narrow range of health consequences, focusing almost exclusively on life expectancy. Some studies have used a cross-sectional design; others have used repeated measures. Richard Wilkinson (1995) ably summarized the state of research in this area as of the mid-1990s: 'A total of at least eight different research workers or groups have reported statistically significant relations between income distribution and measures of mortality using 10 separate sets of data. Of these eight, one has used data exclusively on developing countries, two on a mixture of developed and developing countries, and five exclusively on developed countries. The association has been found to be independent of fertility, maternal literacy, and education in developing countries and [independent] of average incomes, absolute levels of poverty, smoking, racial differences, and various measures of the provision of medical services in developed countries.' A more egalitarian income distribution could add two years to a country's aggregate life expectancy (Wilkinson 1992). More recent work, largely from the United States and to a lesser extent from Canada, has confirmed these findings. For example, Lynch et al. (1998) found that U.S. metropolitan areas with high income inequality and low average income had approximately 140 more deaths per 100,000 residents compared with areas with low income inequality and high income. Their data

Social Cohesion and Health

133

show that, while higher per capita income was associated with lower mortality, this association was weaker than that between income inequality and mortality. Lynch et al. calculated that the magnitude of this difference is comparable to the combined loss of life from lung cancer, diabetes, motor vehicle crashes, human immunodeficiency virus (HIV) infection, suicide, and homicide in 1995. Moreover, preliminary work in Canada, using both aggregate- and individual-level data (Michael Wolfson, Statistics Canada, personal communication) refutes the assertion that such findings could be attributed to a statistical artefact resulting from the use of aggregate-level rather than individual-level data (Gravelle 1998). Several mechanisms through which income inequality may affect health have been posited (see, e.g., Wilkinson 1993; 1994). The first two mechanisms are behavioural: saving on food and other necessities in an effort to maintain socially acceptable standards in more visible areas of consumption (which relates in part to the concept of social exclusion in consumption which Woolley introduces in another chapter in this volume) and engaging in more risky behaviours like smoking, excessive alcohol consumption, or illegal drug use. The other mechanisms are biological. Lynch and colleagues have developed a typology of such mechanisms (John Lynch, University of Michigan, personal communication), which includes materialist, neomaterialist, and psychosocial mechanisms; however, these mechanisms are not necessarily mutually exclusive, and they may operate at different points in the life course. The materialist hypothesis that the very poor cannot afford food and other necessities cannot explain the fact that income and health status follows a gradient with each successively lower income grouping having successively lower health status. There is no absolute level of poverty below which health status suffers and above which health status is unaffected. The neomaterialist mechanism (associated most closely with George Kaplan and John Lynch at the University of Michigan) recognizes the gradient in health status across income levels and posits that individuals at each income level have, for example, less balanced diets, less adequate housing, more exposure to hazardous working conditions or environmental toxins, and less access to social infrastructure, than individuals at the next highest income level (Lynch and Kaplan 1997). The psychosocial mechanism (associated most closely with Richard Wilkinson at the University of Sussex) emphasizes the gradient in health status across income levels and posits that the link between income inequality and health is related to the powerful psychosocial connotations of material differences, not to the material differences per se. This mechanism is really a host

134 John N. Lavis and Gregory L. Stoddart of mechanisms commonly grouped together under the banners of psychoneuroendocrinology or psychoneuroimmunology, fields that are currently being explored extensively (Brunner 1997; Kelly et al. 1997). Psychological processes arising from perceptions of one's status, economic insecurity, or relative deprivation are said to translate into important changes in the nervous, endocrine, and immune systems - changes that, in turn, lead to deteriorations in mental and physical health. In other words, perceptions, feelings, and stimulation are biological events as well as social ones. Examples of the biological impacts of social stress are legion. Consider, for example, the flu that often strikes students after their exams: stress can have a direct impact on the immune system. Similarly, a reduction in immune status associated with bereavement is a well-known phenomenon (Pelletier and Herzing 1988; Evans et al. 1994; Dantzer and Kelley 1989). Now consider what lifelong exposure to constant and relatively high levels of daily stress associated with economic insecurity, relative deprivation, feelings of helplessness, or lack of control over one's life might do! Although much remains to be discovered about the exact nature of the biological pathways linking social and economic environments to ill health, what is now clear is that such pathways do, in fact, exist (Evans et al. 1994; Everson et al. 1997; McEwan 1998). Employment and Working Conditions In response to globalization, trade competition, and technological innovation, labour markets have changed dramatically especially in the past decade, perhaps more than has any other aspect of the social environment. Many workers in advanced economies face limited employment prospects and poor working conditions. The resulting adverse labour market experiences - related either to the availability of work (e.g., unemployment, job insecurity, and overwork) or to the nature of work (e.g., increasing job demands, low job position within a firm, and contingent employment) - can, like income inequality, have profound health consequences. Moreover, the health consequences of these labour market experiences may vary according to the context in which they are experienced. For example, the health consequences of unemployment have been found to vary according to local labour market conditions (e.g., Iversen et al. 1987); they may also vary according to the generosity of unemployment insurance benefits or the degree of reliance on work-related social supports. Such interactions may suggest possible intervention strategies (e.g., changes to benefit levels) or groups at whom interventions should

Social Cohesion and Health

135

be targeted (e.g., groups in particular labour markets or groups reliant on work-related social supports). The literature on the links between labour market experiences and health provides another illustrative example of the influence of the social environment on health. Again, many of the studies have methodological limitations, most significantly the potential for the results to be attributable to differences at baseline (e.g., in health) between those exposed to a labour market experience and those not so exposed, not to the labour market experience per se. Taken together, however, the results are again persuasive. Researchers have examined a range of labour market experiences and their health consequences, which may include behavioural and biological responses, professionally diagnosed disease, and patient-reported mental or physical health and function. Many of these studies have used a cohort design, a more rigorous approach than a cross-sectional design because it permits the examination of whether a labour market experience at one point in time is associated with a health outcome at a later point in time. Only those studies that used a cohort design and for which data on exposure (to a labour market experience) and outcomes could be extracted will be discussed. Experiences that have been examined in such cohort studies include unemployment, job insecurity, job characteristics, job position within the firm, and (in only one study) organizational characteristics of the firm. No cohort studies have examined underemployment or overwork (Lavis 1998). Although the experience of job insecurity has been examined for its associations with a range of health outcomes, other experiences have been more narrowly studied, for example, disease for unemployment in adults, health-related behaviours and mental health and function for unemployment in youth, mostly cardiovascular health outcomes for job characteristics, and mostly cardiovascular system-related individual responses and disease for job position within the firm. Adverse labour market experiences related to the availability of work unemployment and job insecurity - have been consistently found in cohort studies to be associated with negative health outcomes. The absolute risk of a negative health consequence was consistently higher among unemployed adults compared with employed adults. For example, unemployed adults (meaning adults who wish to be employed, but are not) have been found to have increased blood cholesterol levels (Kasl et al. 1968), gains of over 10 per cent in body mass index (Morris et al. 1992), and earlier deaths (Moser et al. 1984; Moser et al. 1986; Iversen et al. 1987; Costa and Segnan 1987; Martikainen 1990; Stefansson 1991; Morris et al. 1994; Martikainen and

136 John N. Lavis and Gregory L. Stoddart Valkonen 1996; Lavis 1997) compared with employed adults. To make this more concrete, Morris et al. (1994) found that the age-adjusted proportion of men who were unemployed (not because of illness) at one point in time and still alive five years later was 2.2 percentage points lower than the percentage of men who were continuously employed and still alive five years later (93.3 per cent compared with 95.7 per cent). One category of exception to the general association between unemployment and negative health outcomes is health-related behaviours like smoking or alcohol consumption (Iversen and Klausen 1986; Kaprio and Koskenvuo 1988; Morris et al. 1992), which could be explained by the relatively high cost of such behaviours for unemployed individuals. The health consequences of unemployment vary according to the local unemployment rate: for example, three studies have found that the mortality rate for unemployed adults was higher in areas (Iversen et al. 1987; Lavis 1997) or periods (Martikainen and Valkonen 1996) with low unemployment rates. The absolute risks of ischemia (i.e., heart trouble) and work absence because of sickness were higher among individuals with insecure jobs compared with individuals with secure jobs (Ferrie et al. 1998; Owens 1966). Adverse labour market experiences related to the nature of work - job characteristics and job position within the firm - have also been found in cohort studies to be associated with a number of negative health outcomes. The absolute risks of a positive coronary heart disease indicator or of mortality within a specified period were higher among individuals with low job decision latitude, low job decision latitude coupled with low job support, or high job strain (i.e., high job demands and low job decision latitude) (Karasek et al. 1982; Astrand et al. 1989; Falk et al. 1992; Alterman et al. 1994). The mortality risks within a specified period were higher for individuals at each successively lower position within a firm (in this case, the U.K.'s civil service) compared with those in the position just above them (Marmot et al. 1978; Rose and Marmot 1981; Marmot et al. 1984). Moreover, a single measure of the social environment - job control - explained more of the variation in health across job positions within a firm than all standard coronary heart disease risk factors taken together (Marmot et al. 1997). The link between labour market experiences and health could follow some of the same mechanisms posited for the relationship between income inequality and health. The first mechanism is again behavioural - engaging in more risky behaviours like smoking, excessive alcohol consumption, or illegal drug use - although the evidence does not support this. The second mechanism is biological: psychological processes that translate into important changes in the nervous, endocrine, and immune systems - changes that, in turn, lead to

Social Cohesion and Health

137

deteriorations in mental and physical health. A third mechanism is not a mechanism per se but an indirect route through other health determinants. Unemployment, for example, could lead to relative poverty and a loss of work-related social supports (i.e., to the social exclusion discussed by Woolley in her chapter), both of which have been found to be associated with ill health. Social Support Increasing life expectancy, larger differentials in life expectancy between husbands and wives, and fewer children to care for elderly parents have meant that more and more elderly individuals lack meaningful amounts of social support. Other factors explain similar deficiencies in other age groups. Social support can provide both a sense of belonging and intimacy and can help people to be more competent and self-efficacious. The literature on the links between social support and health provides a third illustrative example of the influence of the social environment on health. Again, like all research, many of the individual studies have methodological limitations; however, taken as a group, the studies reinforce one another. Because studies have used a range of designs and have examined a range of health consequences, we will concentrate our brief review on cohort studies that examine the association between social support and life expectancy. A review of the eight population-based cohort studies conducted over the past twenty years indicates that people who are isolated have an increased risk of death from a number of causes (Berkman 1995). The association was consistently found in all studies despite differences in the types of communities in which the studies took place, in how social support was measured, in the length of follow-up, and in the factors that the investigators held constant (either by design or by analytic technique). The earliest and best-known study, the Alameda County study (Berkman and Syme 1979), found that men and women who lacked ties to others (through marriage, family, friendships, church, or other groups) were 1.9 to 3.1 times more likely to die in a nineyear follow-up period (from 1965 to 1974) than those who had many more social contacts. As well, more recent studies have found that social support is also associated with survival after a heart attack, suggesting that the influence of social support is not restricted to the onset or progression of clinical disease. As with other aspects of the social environment, social support could affect health through either behavioural or biological mechanisms or both. The behavioural mechanisms could include engaging in more risky behaviours

138 John N. Lavis and Gregory L. Stoddart like smoking (Mermelstein et al. 1986) and excessive alcohol consumption (Berkman and Breslow 1983) or participating in fewer health-related activities like cancer screening (Suarez et al. 1994; Calnan 1985; Kang and Bloom 1993). The biological mechanisms could, again, include changes in the nervous, endocrine, and immune systems. This look at three illustrative pathways through health determinants has a clear conclusion: to the extent that low levels of social cohesion increase income inequality, lead to adverse labour market experiences, or decrease social support, we should expect significant negative health effects in the population. Because the temporal sequence of relationships involving social cohesion and health is critically important to the validity of any conclusions arising from our synthesis of existing empirical research and our analysis using the World Values Survey, we have repeatedly pointed out that health could affect indicators of social cohesion, as well as the reverse. For example, health could affect income-earning potential, with more-healthy people earning more money and less-healthy people earning less. Increasing income inequality could, therefore, lead to a decline in social cohesion. Although our research using the World Values Survey could not distinguish one temporal sequence from another, many studies contained in our synthesis of empirical research were able to make this distinction and their (and our) focus was on health coming at the end of a sequence, not at the beginning. Conclusion This chapter set out to provide some perspective with which to view the relationship between social cohesion and health. In the process, we described new empirical research regarding these relationships in G7 countries, and we synthesized and critically appraised other empirical research to inform discussions about the strength of (some of) these relationships, generally those that involve the determinants of health and specifically those that involve income inequalities, employment and working conditions, and social support. We draw three main conclusions from our endeavour. First, social cohesion can have health consequences. The authors of other chapters in this volume concern themselves primarily with economic consequences and the argument that the relationship between society and the economy is reciprocal (e.g., Helliwell; Stanley). Insofar as the health consequences of social cohesion also have economic consequences, either directly through productivity or through the opportunity costs of health care expenditures (Lavis and Stoddart 1994), we provide further support for this argument. However, we also add a new dimension to the discussion by demon-

Social Cohesion and Health

139

strating that social cohesion can have health consequences which, aside from their economic implications, significantly affect the quality of people's lives. Second, empirical research suggests that these health consequences may be more significant than the average policy maker or citizen (even informed ones) may believe. Note the magnitude of the relationships in our synthesis of empirical research: income distributed more equally could add two years to a country's aggregate life expectancy, or put another way, a more egalitarian income distribution coupled with a higher average income could avert 140 deaths per 100,000 citizens of a given country (Lynch et al. 1998). As well, adults with adverse labour market experiences or limited social supports had higher risks of death than adults with 'good' labour market experiences or social supports. Yet these are only three of many types of health determinants. Some of the other determinants, like early childhood development, also have far-reaching and profound health consequences (see, e.g., Hertzman 1998). Third, the concepts related to social cohesion do not need reconciliation so much as they need links to the right policy environment. Social cohesion may be an all-embracing term, but its dimensions link naturally to specific aspects of government. If one uses Jensen's (1998) typology of five dimensions of paired concepts related to social cohesion, issues related to inclusion or exclusion (like unemployment as discussed in this chapter or social exclusion in consumption as discussed in Woolley's chapter) link naturally to departments like Finance or Human Resources Development, while issues related to legitimacy might link more naturally to departments like Heritage Canada. We did not explicitly link constructs with policy environments, choosing instead (as a first cut) to adopt commonly used indicators of social cohesion: trust to capture one aspect of the belonging/isolation dimension of social cohesion and membership in voluntary organizations to capture the participation/non-involvement dimension. Moreover, our indicators do not explicitly encompass both horizontal associations between people (uniting ties within groups) and vertical associations across divides (bridging ties across groups).8 Our empirical work related to social cohesion can, therefore, only be linked to the policy environment in a general way. We hope that future empirical work will use indicators that encompass both horizontal and vertical ties and will link with specific policy environments. Our review of empirical work related to pathways through health determinants like income distribution or employment/working conditions can be linked to policies produced by specific policy environments like the departments of Finance or Human Resources Development. Our review of empirical work related to social supports can be linked to a range of policy environments, especially for children and the elderly. We believe that this represents a fruitful way forward.

140 John N. Lavis and Gregory L. Stoddart Appendix: Variables Constructed for Analyses Using the World Values Survey Variable

Values

INDEPENDENT VARIABLES (MEASURES OF SOCIAL COHESION)

1

Trust

If respondent answered 'can be trusted' to the question 'generally speaking, would you say that most people can be trusted or you can't be too careful in dealing with people' 0 If respondent answered 'can't be too careful' If missing

Number of memberships in voluntary organizations

#

Membership in one of four categories of voluntary organizations: Religious or church organizations

Sports and cultural organizations

Number of memberships to which the respondent belongs from a list of 16 categories of voluntary organizations, regardless of why s/he chose to belong (range, 0-16) If missing

1

If respondent belongs to religious or church organizations 0 If respondent does not belong to religious or church organizations If missing

1

If respondent belongs to sports and cultural organizations including: (a) social welfare services for elderly, handicapped, or deprived people; (b) education, arts, music, or cultural activities; (c) youth work like scouts, guides, and youth clubs; (d) sports or recreation; (e) voluntary organizations concerned with health; and (f) other groups 0 If respondent does not belong to sports and cultural organizations If missing

Social Cohesion and Health 141 Variable

Values

Political groups

If respondent belongs to political groups including: (a) political parties or groups; (b) local community action on issues like poverty, employment, housing, racial equality; (c) Third world development; (d) conservation, the environment, ecology; (e) professional associations; (f) women's groups; (g) peace movement; and (h) animal rights 0 If respondent does not belong to political groups If missing

Trade union

1 If respondent belongs to a trade union 0 If respondent does not belong to a trade union If missing

1

DEPENDENT VARIABLE

Good healtha

1 If respondent answered very good or good to the question 'all in all, would you describe your state of health these days as very good, good, fair, poor, or very poor' 0 If respondent answered fair, poor, or very poor to the question 'all in all, would you describe your state of health these days as very good, good, fair, poor, or very poor' If missing

POTENTIAL CONFOUNDERS

Age

Young

#

Respondent's age in years If missing

1

If respondent's age between 18 and 44 inclusive

0 If respondent's age 45 or greater If missing Female

1

If respondent female

0 If respondent male If missing White

1

If respondent caucasian/white

0 If respondent negro/black If south asian, east asian, arabic, other, or missing

142 John N. Lavis and Gregory L. Stoddart Variable

Values

Married

1 If respondent married or living as married 0 If respondent divorced, separated, widowed, or single If missing

High income

1 If respondent's income greater than or equal to C$30,000 in Canada in 1981 and in 1990 and greater than or equal to 4, equivalent to latter, in other countries in 1981 and 1990 0 If respondent's income less than C$30,000 in Canada in 1981 and in 1990 and less than 4, which is equivalent to latter, in other countries in 1981 and in 1990 If missing

Years of education

#

Educated

1 If respondent's years of education greater than or equal to 12 years (i.e., high school or more) 0 If respondent's years of education less than 12 years (i.e., less than high school) If missing

High employment grade

1 If respondent performs non-manual work 0 If respondent performs manual work If farmer, agricultural worker, member of armed forces or missing

Employed

1 If respondent employed (i.e., has paid employment) 0 If respondent unemployed If respondent retired/pensioned, housewife, student, other, or missing

a

Age at which respondent completed or will complete education (where 1 if left school at age 12 or less, 2 if 13, and up to 10 if 22 or later) plus 6 If missing

The dichotomization of this categorical variable was conducted using responses to a similar question and scores on the health status index from working age respondents to the National Population Health Survey (Ibrahim, personal communication).

Social Cohesion and Health

143

Notes 1 We would like to thank Heritage Canada for funding this research, Dick Stanley for acting as our contact within the department, and Lars Osberg for prodding us to make sure that our research fit with the broader endeavour that he is coordinating. We would also like to thank Michele Campolieti, formerly a programmer/analyst in the Population/Workforce Studies research program at the Institute for Work & Health, who assisted with the analyses using the World Values Survey. As well, we would like to thank our colleagues in the Population Health Program of the Canadian Institute for Advanced Research, especially Jonathan Lomas and Noralou Roos, for their helpful comments at an early stage in the project, and our colleagues in the Polinomics Group at McMaster University, especially Vandna Bhatia, Stephen Birch, Cathy Charles, Jeremiah Hurley, and Christel Woodward, for their helpful comments on the first draft of the chapter. 2 A fuller treatment of these issues, particularly a discussion about how government performance may be related to both social cohesion and health, can be found in J.N. Lavis and G.L. Stoddart (1999). Social Cohesion and Health. McMaster University Centre for Health Economics and Policy Analysis Working Paper 99-9 (or Canadian Institute for Advanced Research Population Health Program Working Paper PHWP-74). 3 Unfortunately many details of the survey methodology have not been made available to those using the survey (and many researchers have published analyses using the surveys without this information), perhaps because most of the surveys in developed countries were conducted by private polling firms (principally the Gallup chain). Through direct communication with the principal investigator for the Canadian World Values Survey in 1990 (Neil Nevitte 1996), we were able to determine that the response rate for the survey was 62 per cent using a six callback protocol. 4 See web site http://www.worldbank.org/poverty/scapital.htm. 5 To maintain continuity across our analyses, we restricted our list of individuallevel factors to those that have been found to be associated with health - a list that includes age, sex, race, marital status, income, education, employment status, and employment grade. 6 While it may seem odd to use the word 'health' for an outcome category - 'health and function' - which is meant to define the concept of health, this is the label used in the Evans and Stoddart (1990) framework. Its meaning is explained below. 7 Japan has a very high level of health status compared to other developed countries when the measure of health status is life expectancy (OECD 1998). This difference between objective measures and subjective responses may be explained, at least in part, by cultural biases. In discussing self-reported health status using a slightly different set of responses, Wagner et al. (1998) state: The response choices

144 John N. Lavis and Gregory L. Stoddart (excellent, very good, good, fair, and poor) were the subject of much discussion, particularly how appropriate the attribute excellent is when talking about health. In Germany and France, for example, qualifying one's health as excellent may be viewed as "tempting one's fate," whereas in Japan, it is socially desirable to be modest and thus not to speak about one's health as excellent.' 8 Our finding that older, white, high-income, educated, and employed individuals were more trusting and had a larger number of memberships in voluntary organizations than younger, black, low-income, less well educated, and unemployed individuals reinforces the need to examine ties across groups.

References Almond, G.A., and S. Verba. 1963. The Civic Culture: Political Attitudes and Democracy in Five Nations. Princeton, NJ: Princeton University Press. Alterman, T., R.B. Shekelle, S.W. Vernon, and K.D. Burau. 1994. 'Decision Latitude, Psychological Demand, Job Strain, and Coronary Heart Disease in the Western Electric Study.' American Journal of Epidemiology 139(6): 620-7. Amick, III B., S. Levine, A.R. Tarlov, and Walsh D. Chapman, eds. 1995. Society and Health. New York: Oxford University Press. Astrand, N.E., B.S. Hanson, and S.O. Isacsson. 1989, 'Job Demands, Job Decision Latitude, Job Support, and Social Network Factors as Predictors of Mortality in a Swedish Pulp and Paper Company.' British Journal of Industrial Medicine 46: 334-40. Berkman, L.F. 1995. 'The Role of Social Relations in Health Promotion.' Psychosomatic Medicine 57: 245-4. Berkman, L.F., and L. Breslow. 1983. Health and Ways of Living. New York: Oxford University Press. Berkman, L.F., and S.L. Syme. 1979. 'Social Networks, Host Resistance and Mortality: A Nine-Year Follow-Up Study of Alameda County Residents.' American Journal of Epidemiology 109: 186-204. Blane, D., E. Brunner, and R. Wilkinson. 1996. Health and Social Organization: Towards a Health Policy for the 21st Century. New York, NY: Routledge. Bruhn, J., and S. Wolf. 1979. The Roseto Story: An Anatomy of Health. Norman, Okla: Oklahoma University Press. Brunner, E. 1997. 'Stress and the Biology of Inequality.' British Medical Journal 314: 1472-6. Calnan, M. 1985. 'Patterns in Preventive Behavior: A Study of Women in Middle Age.' Social Science and Medicine 20: 263-8. Costa, G., and N. Segnan. 1987. 'Unemployment and Mortality.' British Medical Journal 294: 1550-1.

Social Cohesion and Health 145 Dantzer, R., and K.W. Kelley. 1989. Stress and Immunity: 'An Integrated View of Relationships between the Brain and the Immune System.' Life Sciences 44: 19952008. Egolf, B., J. Lasker, S. Wolf, and L. Potvin. 1992. 'The Roseto Effect: A 50-Year Comparison of Mortality Rates.' American Journal of Public Health 82: 1089-92. Evans, R.G., and G.L. Stoddart. 1990. 'Producing Health, Consuming Health Care.' Social Science and Medicine 31(12): 1347-63. Evans, R.G., M.L. Barer, and T.R. Marmot. 1994. Why Are Some People Healthy and Others Not: The Determinants of Health of Populations. Hawthorne, NY: Aldine de Gruyter. Evans, R.G., M. Hodge, and I.E. Pless. 1994. 'If Not Genetics, Then What? Biological Pathways and Population Health.' In R.G. Evans, M.L. Barer, and T.R. Marmor, eds., Why Are Some People Healthy and Others Not? The Determinants of Health in Populations. Hawthorne, NY: Aldine de Gruyter. Everson, S.A., G.A. Kaplan, D.E. Goldberg, R. Salonen, and J.T. Salonen. 1997. 'Hopelessness and 4-Year Progression of Carotid Atherosclerosis: The Kuopio Ischemic Heart Disease Risk Factor Study.' Arteriosclerosis Thrombosis and Vascular Biology 17: 1490-5. Falk, A., B.S. Hanson, S.O. Isacsson, and P.O. Ostergren. 1992. 'Job Strain and Mortality in Elderly Men: Social Network, Support, and Influence as Buffers.' American Journal of Public Health 82: 1136-9. Federal/Provincial/Territorial Advisory Committee on Population Health. 1994. Strategies for Population Health: Investing in the Health of Canadians. Ottawa: Minister of Supply and Services Canada. Ferric, I.E., M.J. Shipley, M.G. Marmot, S. Stansfeld, and G. Davey Smith. 1998. The Health Effects of Major Organizational Change and Job Insecurity.' Social Science and Medicine 46(2): 243-54. Frank, J.W. 1995. 'The Determinants of Health: A New Synthesis.' Current Issues in Public Health 1:233-40. Fylkesnes, K. 1993. 'Determinants of Health Care Utilization - Visits and Referrals.' Scandinavian Journal of Social Medicine 21: 40-50. Goldstein, M.S., J.M. Siegel, and R. Boyer. 1984. 'Predicting Changes in Perceived Health Status.' American Journal of Public Health 74: 611-14. Gravelle, H. 1998. 'How Much of the Relation between Population Mortality and Unequal Distribution of Income is a Statistical Artefact?' British Medical Journal 316:382-5. Helliwell, J.F. 1996. Do Borders Matter for Social Capital? Economic Growth and Civic Culture in U.S. States and Canadian Provinces. Cambridge, Mass.: National Bureau of Economic Research.

146 John N. Lavis and Gregory L. Stoddart Hertzman, C. 1994. 'The Lifelong Impact of Childhood Experiences: A Population Health Perspective.' Daedalus 123(4): 167-80. Hertzman, C. 1998. The case for Child Development as a Determinant of Health.' Canadian Journal of Public Health 89 (Supplt): S14-S19. Hulka, B.S., and J.R. Wheat. 1985. 'Patterns of Utilization: The Patient Perspective.' Medical Care 23: 438-60. Idler, E.L., and Y. Benyamini. 1997. 'Self-Rated Health and Mortality: A Review of Twenty-Seven Community Studies.' Journal of Health and Social Behaviour 38: 21-37. Inglehart, R.F., N. Nevitte, and M. Basanez. 1996. The North American Trajectory: Cultural, Economic and Political Ties among the United States, Canada, and Mexico. New York: Aldine de Gruyter. Iversen, L., O. Andersen, P.K. Andersen, K. Christoffersen, and N. Keiding. 1987. 'Unemployment and Mortality in Denmark, 1970-80.' British Medical Journal 295: 879-84. Iversen, L., and H. Klausen. 1986. 'Alcohol Consumption among Laid-Off Workers before and after Closure of a Danish Ship-Yard: A 2-Year Follow-Up Study.' Social Science and Medicine 22(1): 107-9. Jenson, J. 1998. Mapping Social Cohesion: The State of Canadian Research. Ottawa: Canadian Policy Research Networks. Judge, K. 1995. 'Income Distribution and Life Expectancy: A Critical Appraisal.' British Medical Journal 311: 1282-5. Kang, S.H., and J.R. Bloom. 1993. 'Social Support and Cancer Screening among Older Black Americans.' Journal of the National Cancer Institute 85: 737-42. Kaprio, J., and M. Koskenvuo. 1988. 'A Prospective Study of Psychological and Socioeconomic Characteristics, Health Behavior and Morbidity in Cigarette Smokers prior to Quitting Compared to Persistent Smokers and Non-Smokers.' Journal of Clinical Epidemiology 41(2): 139-50. Karasek, R.A., T. Theorell, J.E. Schwartz, C.F. Pieper, and L. Alfredsson. 1982. 'Job, Psychological Factors and Coronary Heart Disease.' Advances in Cardiology 29: 62-7. Kasl, S.V., S. Cobb, and G.W. Brooks. 1968. 'Changes in Serum Uric Acid and Cholesterol Levels in Men Undergoing Job Loss.' Journal of the American Medical Association 206(7): 1500-7. Kawachi, I. 1998. 'Social Cohesion and Health.' Paper prepared for the Kansas Conference on Health and Its Determinants, Wichita, Kansas, 20-1 April. Kawachi, I., B.P. Kennedy, K. Lochner, and D. Prothrow-Stith. 1997. 'Social Capital, Income Inequality, and Mortality.' American Journal of Public Health 87: 1491-8.

Social Cohesion and Health

147

Kelly, S., C. Hertzman, and M. Daniels. 1997. 'Searching for the Biological Pathways between Stress and Health.' Annual Review of Public Health 18: 43762. Krakau, I. 1991. 'Perception of Health and Use of Health Care Services in a Swedish Primary Care District: A Ten Years' Perspective.' Scandinavian Journal of Primary Health Care 9: 103-8. Lavis, J.N. 1997. An Inquiry into the Links between Labour-Market Experiences and Health. Doctoral dissertation. Harvard University, Cambridge, Mass. - 1998. 'Labour-Market Experiences and Health: A Systematic Review of Cohort Studies.' In J.E. Hurley, C.A. Charles, J.N. Lavis, S. Birch, G.M. Torrance, J. Lomas, M. Giacomini, V. Bhatia, G.L. Stoddart, and V. Vincent, Health Policy in the Era of Population Health: An Exploration of Changing Roles. Report submitted to Health Canada through the National Health Research and Development Program. Lavis, J.N., and G.L. Stoddart. 1994. 'Can We Have Too Much Health Care?' Daedalus 123(4): 43-60. Lundberg, O., and K. Manderbacka. 1996. 'Assessing Reliability of a Measure of Self-Rated Health.' Scandinavian Journal of Social Medicine 24: 218-24. Lynch, J.W., and G.A. Kaplan. 1997. 'Understanding How Inequality in the Distribution of Income Affects Health.' Journal of Health Psychology 2(3): 297-314. Lynch, J.W., G.A. Kaplan, and S.J. Shema. 1997. 'Cumulative Impact of Sustained Economic Hardship on Physical, Cognitive, Psychological, and Social Functioning.' New England Journal of Medicine 37(26): 1889-95. Lynch, J.W., G.A. Kaplan, E.R. Pamuk, R.D. Cohen, K.E. Heck, J.L. Balfour, and I.H. Yen. 1998. 'Income Inequality and Mortality in Metropolitan Areas of the United States.' American Journal of Public Health 88: 1074-80. Marmot, M.G., M.J. Shipley, and G. Rose. 1984. 'Inequalities in Death - Specific Explanations of a General Pattern?' Lancet (5 May): 1003-6. Marmot, M.G., G. Rose, M.J. Shipley, and P.J.S. Hamilton. 1978. 'Employment Grade and Coronary Heart Disease in British Civil Servants.' Journal of Epidemiology and Community Health 32: 244-9. Marmot, M.R., H. Bosma, H. Hemingway, E. Brunner, and S. Stansfeld. 1997. 'Contribution of Job Control and Other Risk Factors to Social Variations in Coronary Heart Disease Incidence.' Lancet 350: 235-9. Martikainen, P.T. 1990. 'Unemployment and Mortality among Finnish Men, 19815.' British Medical Journal 301: 407-11. Martikainen, P.T., E. Lahelma, A. Aromaa, and the Mini-Finland Research Group. 1996. 'Perceived Health and Mortality: A Follow-Up Study.' Paper presented at the 6th Biennial Conference of the European Society of Health and Medical

148 John N. Lavis and Gregory L. Stoddart Sociology, Health and Social Change in the Integration of Europe. Budapest, Hungary, 29-31 Aug. Martikainen, P.T., and T. Valkonen. 1996. 'Excess Mortality of Unemployed Men and Women during a Period of Rapidly Increasing Unemployment.' Lancet 348: 909-12. Maxwell, J. 1996. Social Dimensions of Economic Growth. Eric John Hanson Memorial Lecture Series, vol. 8. Edmonton: University of Alberta. McEwan, B.S. 1998. 'Protective and Damaging Effects of Stress Mediators.' New England Journal of Medicine 338: 171-9. Mermelstein, R., S. Cohen, E. Lichtenstein, J.S. Baer, and T. Kamarck. 1986. 'Social Support and Smoking Cessation and Maintenance.' Journal of Consulting Clinical Psychology 54: 447-53. Morris, J.K., D.G. Cook, and A.G. Shaper. 1992. 'Non-employment and Changes in Smoking, Drinking and Body Weight.' British Medical Journal 304: 536-41. Morris, J.K., D.G. Cook, and A.G. Shaper. 1994. 'Loss of Employment and Mortality.' British Medical Journal 308: 1135-9. Moser, K.A., A.J. Fox, and D.R. Jones. 1984. 'Unemployment and Mortality in the OPCS Longitudinal Study.' Lancet (2): 1324-9. Moser, K.A., A.J. Fox, D.R. Jones, and P.O. Goldblatt. 1986. 'Unemployment and Mortality: Further Evidence from the OPCS Longitudinal Study 1971-1981.' Lancet (1): 365-7. Nevitte, N. 1996. The Decline of Deference: Canadian Value Change in CrossNational Perspective. Peterborough, Ont.: Broadview Press. Noack, R.H. 1997. Research for Health Promotion: A Challenge for the 21st Century. Report prepared for the Advisory Committee on Health Research for the World Health Organization. Graz, Austria: University of Graz, Institute of Social Medicine. OECD. 1998. OECD Health Data 98. Paris: OECD. Osberg, L. 1998. 'Economic Policy Variables and Population Health.' National Forum on Health, Canada Health Action: Building on the Legacy - Determinants of Health: Settings and Issues. Ste-Foy, Que.: Editions Multi Mondes, 531-65. Owens, A.C. 1966. 'Sick Leave among Railwaymen Threatened by Redundancy: A Pilot Study.' Occupational Psychology 40: 43-52. Pelletier, K.R., and D.L. Herzing. 1988. 'Psychoneuroimmunology: Toward a Mindbody Model.' Advances 5 (1): 27-56. Pope, G.C. 1988. 'Medical Conditions, Health Status, and Health Services Utilization.' Health Services Research 22: 857-77. Rose, G., and M.G. Marmot. 1981. 'Social Class and Coronary Heart Disease.' British Heart Journal 45: 13-19.

Social Cohesion and Health

149

Rowan, K. 1994. 'Global Questions and Scores.' In C. Jenkinson, ed., Measuring Health and Medical Outcomes. London: UCL Press, 54-67. Sampson, R.J., S.W. Raudenbush, and F. Earls. 1997. 'Neighbourhoods and Violent Crime: A Multilevel Study of Collective Efficacy.' Science 277: 918-24. Stefansson, C.G. 1991. 'Long-Term Unemployment and Mortality in Sweden, 19801986.' Social Science and Medicine 32(4): 419-23. Stoddart, G.L. 1995. The Challenge of Producing Health in Modern Economies. (Working Paper 95-15). Toronto: Canadian Institute for Advanced Research Population Health Program. Stone, D. 1997. Policy Paradox: The Art of Political Decision Making. New York: W.W. Norton. Suarez, L., L. Lloyd, N. Weiss, T. Rainbolt, and L. Pulley. 1994. 'Effect of Social Networks on Cancer-Screening Behavior of Older Mexican American Women.' Journal of the National Cancer Institute 86: 775-9. Wagner, A.K., B. Gandek, N.K. Aaronson, et al. 1998. 'Cross-Cultural Comparisons of the Content of SF-36 Translations across 10 Countries: Results from the IQOLA Project.' Journal of Clinical Epidemiology 51(11): 925-32. Wilkinson, R.G. 1992. 'Income Distribution and Life Expectancy.' British Medical Journal 304: 165-8. - 1993. The Epidemiological Transition: From Material Scarcity to Social Disadvantage?' Proceedings of the 11th Honda Foundation Discoveries Symposium: Prosperity, Health and Well-Being, Toronto, 16-18 October, 193-207. - 1994. 'The Epidemiological Transition: From Material Scarcity to Social Disadvantage? Daedalus 123(4): 61-77. - 1995. 'Commentary: A Reply to Ken Judge: Mistaken Criticisms Ignore Overwhelming Evidence.' British Medical Journal 311: 1285-7. Wolf, S., and J. Bruhn. 1992. The Power of Clan: A 25-Year Prospective Study of Roseto, PA. New Brunswick, NJ: Transaction Publishers. Wolfson, M.C., G. Kaplan, J. Lynch, N.A. Ross, and E. Backlund. 1999. The Relationship between Income Inequality and Mortality Is Not a Statistical Artefact: An Empirical Assessment.' British Medical Journal 319: 953-7. World Health Organization. 1986. Ottawa Charter for Health Promotion. Ottawa: World Health Organization.

Social Cohesion and Voluntary Activity: Making Connections FRANCES WOOLLEY1

According to the Organisation for Economic Cooperation and Development, 'A strong social fabric provides a secure basis for the flexibility and risktaking which are the life-blood of vibrant economic activity and wealth creation' (OECD 1997: 7). Voluntary activity forms part of this social fabric. As Dick Stanley argues, social cohesiveness 'manifests itself in the existence of voluntary relationships.'2 Voluntary action is thought to have both internal effects on members, teaching them to be more cooperative, and external effects on the wider polity, fostering social cooperation (Jenson 1999: 11). If this is true, then structural adjustment policies may create a double dividend. For example, the downsizing of government might not only lessen the burden of the state on the economy, but to the extent that the space created by government downsizing is filled with voluntary activity, social cohesion might grow (OECD 1997: 22). Moreover, the connection between social cohesion and the voluntary sector suggests a new role for government activity: Possibly governments can make small policy interventions which, by encouraging voluntary activities, have large dividends in terms of social cohesiveness and, in turn, economic and human prosperity. If policy makers are to build social cohesiveness by encouraging voluntary activity, much more needs to be known about social cohesion, volunteering, and the connection between the two. What fosters voluntarism? What hinders it? How can voluntarism both create, or be created by, social cohesiveness? Social cohesion and voluntary activity are both multidimensional. This makes the relationship between the two complex and, likely, non-linear. Heritage Canada describes social cohesion as 'the ongoing process of developing a community of shared values, shared challenges and equal opportunity within Canada, based on a sense of trust, hope and reciprocity among all Canadians' (Strategic Research and Analysis Directorate 1997: 4 emphasis added). Jane Jenson (1999) argues that social cohesion has five dimensions: belonging/

Social Cohesion and Voluntary Activity

151

isolation, inclusion/exclusion, participation/non-involvement, recognition/rejection, legitimacy/illegitimacy. The word voluntary is defined by the Concise Oxford Dictionary (Allen 1990) as: voluntary (1) done, given, or acting of one's own free will; [physiology] under the conscious control of the brain (2) working or done without payment. Volunteer (1) a person who freely offers to do something - a person who freely enrols for military service rather than be conscripted (2) a person who works for an organization without being paid. The focus in this chapter is primarily on the second of these two meanings: voluntary activity as characterized by the absence of direct, financial exchanges. The term 'volunteering' is used here to refer to gifts of time or unpaid work, and the term 'donations' is used to refer to financial or material gifts. This chapter distinguishes between 'volunteering,' which is the giving of time and skills to groups and organizations, and 'helping out,' which is providing direct, informal, help to friends and neighbours. Statistics Canada's National Survey of Giving, Volunteering and Participating (NSGVP) makes this same distinction in the same way. Volunteering is not just about giving time and assistance to others, but about doing so in a particular way (unpaid) within a particular institutional context (voluntary or non-profit groups and associations). Figure 1 illustrates the relationship between voluntary and other forms of activity. Social needs are met through unpaid work (the upper portion of the diagram) or paid work (the lower portion). Each type of work can occur in either the public or the private sphere. Voluntary activity is unpaid work in the public sphere. Unpaid work in the private sphere takes place in families, between friends, and among neighbours. Paid work in the public sphere includes governments and cooperatives. Paid work in the private sphere is the private sector of profit-maximizing firms. Although volunteering may seem, from the above discussion, to be a clear and well-understood concept, it is extremely problematical for economists. Economists are accustomed to explaining people's behaviour through the prism of utility maximization: A person will take an action if the benefits of doing so outweigh the costs. This creates a paradoxical situation. If volunteering is taken to mean unpaid activity, why would people choose to do it? How can the benefits of unpaid activity possibly outweigh the costs? Explanations of Voluntary Activity Why do some activities fall into the 'voluntary' portion of Figure 1, instead of being done by government, family, or the private sector? What factors facilitate, or hinder, the growth of the voluntary sector, and the link between

152 Frances Woolley Figure 1 A typology of work

voluntary activity and social cohesion? Will a shrinking of the government quadrant (in Figure 1) lead to growth in the other quadrants, or will it simply result in needs going unmet? The simplest reason for people to volunteer is to provide, through their own actions, those goods not provided by other social or economic institutions. For example, although private firms may find it unprofitable to provide local public goods, an individual might volunteer to maintain the local skating rink because she enjoys using the rink herself, or coach a children's soccer team so that her children have a team to play on. This explanation uses what is called a private provision of public goods model because people are volunteering to provide, privately, goods and services. These people are not motivated by concern for others, but only by their own self-interest. The private provision of public goods model is a useful starting point; however, it does not fit some facts about voluntary activity. For example, the model predicts that, as the economy becomes extremely large, the amount of public goods provided voluntarily will tend towards zero (Andreoni 1988), as people free ride on the contributions of others. Similarly, government provision is predicted to crowd out voluntary provision dollar for dollar or one for one. Yet modern economies with large government sectors do have strong voluntary sectors, and this is contrary to what the model predicts. .

Social Cohesion and Voluntary Activity 153 There are various ways of modifying the simple private provision model. Andreoni (1995) suggests that people get a 'warm glow' from giving. A sense of duty, concern for others, or a desire for social status are all reasons for people to volunteer, even if other people or governments are already providing public goods. Robert Sugden (1984) introduces the concept of 'reciprocity' to explain voluntary activity. Reciprocity is a norm which holds that, if other people contribute, you are morally obligated not to free ride on, or take advantage of, their contributions. If everyone has the same incomes and taste for the public good, reciprocity can lead to an efficient level of public goods provision. But how are norms of reciprocity created and sustained? One way of enforcing reciprocity is by forming a 'club,' which excludes people who do not contribute to the good. Sports and recreational associations, professional associations, trade unions, cultural organizations such as local theatres, and perhaps even churches can be seen as clubs in this sense. Clubs can provide goods that markets or governments do not, but those goods may be for the enjoyment of members only. A final motivation for voluntary activity is as a type of work that is not paid, but leads to future employment or provides on-the-job training. The employer saves a regular starting salary and gets an opportunity to observe a worker's productivity before actually hiring him or her. Volunteer workers get a chance for eventual employment. Such workers are particularly important in health, education, and social services, as well as in the recreation, community services, and cultural sectors. Kathleen Day and Rose Anne Devlin (1998) have concluded that volunteering increases earnings by approximately 7 per cent. However, while volunteering for sports organizations and other (a residual category) organizations generates a significant positive earnings premium, volunteering for religious organizations carries a statistically significant earnings penalty. Government Spending If people volunteer simply to provide goods and services, then government spending would be expected to crowd out voluntary activity. The idea that voluntary activity can and will replace government provision is embraced by, for example, the OECD: 'Governments will increasingly stop doing directly what may be delegated to regional, municipal, co-operative, mutualist or private organisations' (OECD 1997: 73). This will mean 'letting citizens themselves - and no longer cold and distant bureaucracies - devise "neighbourhood" solutions for everything pertaining to education, the fight against pov-

154 Frances Woolley erty, exclusion, drug trades, taking care of the young, dynamising a local environment, etc' (OECD 1997: 74). Do the facts support the theory? Comparing across countries, there is a positive relationship between public expenditure as a percentage of gross domestic product (GDP) and people's membership in voluntary organizations (Putnam 1995), partly because of high membership and high government spending in the Nordic countries (Norway, Sweden, Denmark, and Finland). Although government spending crowds out voluntary provision to some extent, it is only at levels of between 5 and 28 per cent - much less than the 100 per cent crowding out suggested by theoretical models (e.g., Ferris and West 1998: 15). The degree of crowding out also appears to depend on the type of government expenditures. Day and Devlin (1996) have found, for example, that in Canada the number of volunteers was negatively related to government expenditures on economyrelated activities, but positively related to government expenditures on health care. Meg Luxton (1998: 65) cites similar findings: 'British studies indicate that people are most likely to give support and aid to others when the others already have some sources of support and the caregivers are not faced with the possibility of having to take on full responsibility for the other.' More fundamentally, both volunteerism and government activity grow in similar social climates. Income inequality can make collective action difficult, whether through governments or voluntary organizations, while strong communities, religious values, and certain family structures promote collective action by one means or the other. Income Inequality Greater income inequality can make collective action more difficult, whether through governments or voluntary associations. When top income earners become relatively better off, they pay more income tax both in absolute terms and as a percentage of their income. This increases their 'tax price' for a state-provided consumer service (such as public health care). Although the average voter may still support public health care, the affluent have incentives to purchase political influence and push for lower taxes, thereby undermining the democratic process. When it comes to voluntary action, people with similar incomes will find it easier to cooperate. When people have different incomes, even though everyone may be made better off by cooperating, some may benefit more than others, depending on how the benefits and costs of cooperation are shared. Voluntary associations based on norms of reciprocity will be more effective when people are in similar situations (Sugden 1984). If the community skat-

Social Cohesion and Voluntary Activity 155 ing rink is maintained by volunteers, those who do not skate, or who use a private rink, will have no wish to maintain the local community rink. By not reciprocating the contribution of others, these non-volunteers erode the norm of reciprocity. Moreover, because it is harder for heterogeneous communities to agree on a fair division of costs and benefits, cooperation will be hard to sustain. If voluntary activity has more to do with gaining work experience than with providing public goods and services, it will rise when the opportunity cost of time falls. There was a 15 per cent increase in the number of youth (15- to 24-year-olds) volunteering in Canada between 1987 and 1997, a period that coincided with decreasing employment opportunities and lower wages for young Canadians. When asked at the time why they volunteer, 54 per cent of youth said that they want 'to improve job opportunities,' giving direct support to the theory of volunteering as gateway to future employment, while 68 per cent said they volunteer to 'explore their own abilities,' and 82 per cent to 'use their skills and abilities,' supporting the idea that volunteering is a way for youth to acquire valuable human capital (more than one answer per respondent was allowed; Statistics Canada 1998). For these people, volunteering may simply be employment at a wage of zero. The empirical evidence from the Luxembourg Income Study (LIS) and the World Values Survey (WVS), however, indicates that greater income inequality is associated with lower levels of voluntary activity. Voluntarism was calculated from the WVS and measured using the percentage of those surveyed belonging to or doing unpaid work for a voluntary organization. Income inequality was measured using the Gini coefficient, with values calculated from the LIS by Gottschalk and Smeeding (1997). Figures in both cases are for 1991 or 1992.3 A simple analysis of income inequality and voluntarism finds a strong negative relationship (r = -.47). It is extremely optimistic to think that the voluntary sector can offset rising income inequality. Instead, available evidence provides greater support to the view that rising income inequality erodes the foundation for voluntarism. Protestantism Is the inverse relationship between income inequality and voluntary activity simply a coincidence? Perhaps it is the 'broader communitarian sociability,' to use an OECD (1997) term, of Northern European countries that creates democratic governance, lower income inequality, and a commitment to voluntarism. Many things about these countries - religion, family structure, climate, or some other factor - could create this particular pattern. The data

156 Frances Woolley Figure 2

Protestantism and rate of belonging, taken from 1991 World Values Survey.

indicate that the countries with the highest levels of voluntary activity are all predominantly Protestant. Why might there be this connection? Figure 2 shows rates of volunteering and religious beliefs for a number of countries in Western Europe and North America taken from the 1991 WVS. In countries with significant Protestant populations, more people belong to or do unpaid work for voluntary organizations. For the countries studied, the correlation between the voluntarism rate and the percentage of the population that is Protestant is strong, positive, and statistically significant (r = 0.76). Why do predominately Protestant societies have so much more voluntary activity? One possible explanation centres around the internal institutional structure of the church. As Robert Putnam (1993: 107) puts it, in the Roman Catholic Church, 'vertical bonds of authority are more characteristic ... than horizontal bonds of fellowship.' Orthodox Christian and Moslem churches also tend to have hierarchical structures (Knack and Keefer 1997: 1283).4 By way of contrast, in most Protestant religions, local churches are self-governing. Local governance tends to create small congregations, more churches, and a corresponding need for voluntary activity, for tasks ranging from parish administration to building maintenance. Voluntary activity is unpaid work taking place within a specific institutional context, that of the voluntary association. Conditions that lead to the growth of many small, non-hierarchical organizations will provide a warm climate for voluntarism. A second set of explanations centre around the social norms of religions. In general, the observant tend to volunteer more. To the extent that religion

Social Cohesion and Voluntary Activity 157 tells us to love our neighbours as ourselves, praises the virtue of charity, and restrains profit motives (e.g., through prohibitions on usury), it may increase people's commitment to give to others. Francois Vaillancourt (1994) has found that Protestants volunteer more than Catholics do. What is special about Protestant social norms? A key Protestant principle is the 'believers' church,' (Weber 1946: 313), that is, only people who choose, who volunteer their belief, should belong to the church. Catholic churches, by way of contrast, are just that - catholic, universal, all-embracing. For Protestants, for whom church membership is a matter of choice, of personal belief, the ethic of voluntarism may spill over from religious life to all life. If for Protestants, religion is a matter of personal, individual belief, there may be greater room for divergence in interpretation of the faith - and tolerance of non-believers. If something intrinsic to the Protestant belief system has led to the emergence of strong voluntary associations in Northern Europe and North America, then it would be unrealistic to expect these institutions to emerge rapidly in other countries. Creating a particular form of civil society may involve fundamentally changing people's beliefs and values, including those religious beliefs that are at the heart of people's sense of self. Moreover, if high rates of voluntarism are part of a specifically Protestant ethic, one should be cautious about seeing voluntarism as a universal solution to social problems. Attempting to impose a particular set of values on others not only shows a lack of respect and tolerance; it is also doomed to failure. Pluralism Alternatively, voluntarism may flourish in Protestant countries because of the number and variety of churches, not because of any particular elements of Protestant theology. The universality of the Catholic Church gives it, in Catholic societies, a social monopoly. A universal church can press for the creation of other social institutions - church or government - to do what is accomplished through voluntary organizations in religiously diverse countries. Examples of Catholic Church involvement in social movements abound in Europe: the Solidarity trade union in Poland or Christian Democratic parties taking policy initiatives in response to papal encyclicals in Western Europe. Catholic influence over social policy can be found in reproductive or family legislation in Ireland and in the former West Germany. Quebec, too, has spearheaded social initiatives, such as child care at $5 a day, which would be unthinkable in many other Canadian provinces. Greater social solidarity in societies that have a Catholic majority may perhaps make greater use of the state possible as a mechanism for ensuring social order and social justice.

158 Frances Woolley The number of and variety of churches, as opposed to their theological content, is emphasized in recent research on the economics of religion: 'Every available measure of piety ... is greater in countries with numerous competing churches (lannaccone 1998: 1486). Less or no significance is attached to a religion's principles. lannaccone argues, 'People's religious affiliation or degree of religiosity seems not to influence their attitudes concerning capitalism, socialism, income redistribution, private property, free trade, and government regulation ... This lack of correlation between religious and economic thinking is, of course, just one more blow to Weber's "Protestant Ethic" thesis' (1998: 1477-8). The idea that religious competition matters is supported by the fact that there are higher levels of voluntary activity in countries such as the United States (where, according to the WVS), 70.5 per cent of people belong to at least one voluntary organization, Canada (where 63.6 per cent belong), and the Netherlands (with 84.4 per cent) than in countries such as Great Britain (where 53.1 per cent belong). Although Britain has a greater percentage of Protestants, it also has an established (Anglican) church. Whatever the other advantages of social competition, successful communities will be, along at least some dimensions, homogeneous communities. People choose where they live, and they often choose to live with people similar to themselves. As Jane Friesen shows (in her paper in this volume), local communities have relatively little divergence in terms of income, and, often, other attributes, as well. A fragmented society may still have many small, homogeneous, internally cohesive communities. A strong voluntary sector can, therefore, exist at the local level, even when collective action at the national or provincial level is impossible. In this way, pluralism may strengthen civil society, as measured by within-group cohesion, even if cohesion across groups falls. It is perhaps for this reason that the OECD predicts that citizens can devise 'neighbourhood' solutions for the fight against poverty, exclusion, drug trades, taking care of the young, or dynamizing a local environment (OECD 1997: 74). Voluntary action can be effective at the neighbourhood level, in part, because neighbourhoods are homogeneous. Can homogeneous neighbourhoods take effective action against social exclusion? Strength of Family and Other Institutions The Protestant, high-volunteerism countries of Europe and North America have another common feature: Household units in these countries tend to be small, based on nuclear-type families, rather than larger units based on com-

Social Cohesion and Voluntary Activity 159 munity-type families (Todd 1985). Kinship networks may be large, but they are not very important in mediating economic transactions. The relationship between government and voluntary provision has been the subject of much research. In contrast, the relationship between voluntary provision and other non-market institutions, such as family supports, has been almost completely ignored. Most unpaid work is not considered voluntary activity. Caring for children, elderly dependants, or other household members is not counted as voluntary work in official Canadian statistics. However, it is much more common. In 1997, the National Survey of Giving, Volunteering and Participating (NSGVP) found that three Canadians in ten were volunteers, but seven in ten Canadians provided some other form of care, for example, help to relatives not living with them, or help to others than relatives, such as helping out with shopping, babysitting, or driving (Statistics Canada 1998). Put another way, of the 216 minutes a day that the average Canadian spent in unpaid work (according to the 1992 General Social Survey), only 23 minutes involved civic and voluntary activity.5 Recognizing that voluntary activity is only one part of the social support network is crucial to understanding the relationship between social cohesion and voluntary activity. Arguably, one reason for the weak relationship found in empirical research between the level of government intervention or activity and the level of voluntary activity is that the studies have failed to control for the underlying social structures that influence both government and voluntary sectors. Todd (1985) has argued that societies with smaller, more fragmented family units tend to create strong non-family institutions, including both social democratic governments and the voluntary sector, which are unnecessary in societies with other family structures. Strong commitments supported by strong social, moral, and legal sanctions are duties or responsibilities, not voluntary activity. Voluntary activities fall between duties, on the one hand, and purely self-interested behaviour, on the other. Voluntary activity can be seen as a sign of both cohesion and disintegration. Without some cohesion, some sense of shared values or commitments, people would not volunteer at all - but in a truly cohesive society, there may not be much volunteer work because obligations to others are seen as responsibilities that must be fulfilled, not something that is chosen. For example, we have volunteer services like Meals on Wheels in Canada in part because, generally, the elderly live on their own instead of in extended families, and many do not have family available to care for them. This raises another issue. The model of civil society that rests on voluntarism, and that was developed in the Anglo-Saxon countries (OECD 1997: 74) will not obviously be readily adopted by those with different family structures. Recent

160 Frances Woolley immigrants to Canada have different family structures from native-born Canadians. Older immigrant women, for example, are more likely to live with extended family than are other older Canadians (Woolley 1998). When family substitutes for voluntary association, the amount of voluntary activity observed will go down, even if there is no decrease in helping others out. Substitution of family for civic organization may have no implications for the amount of social cohesiveness, but it may change the location of social networks and the form of social interaction. The voluntary sector is only one of many social institutions. The strength of the voluntary sector may indicate either a healthy society with many strong institutions or instability and contradictions within other social institutions. The strength of the voluntary sector needs to be assessed relative to the demands on it. Does Voluntary Activity Build Social Cohesion? A key premise of the social cohesion policy agenda is that voluntary activity builds social cohesion. Here let us consider two paths from volunteering to cohesion: giving and participating. The idea that giving strengthens social connections dates back to Titmuss's (1971) analysis of blood donation, in his book entitled The Gift Relationship. Titmuss argued that altruistic acts, such as giving blood, strengthen social relations. Social interaction is irrelevant; indeed, blood donation is taken to strengthen social relations despite, or even because of, the anonymity of donor and recipient. In contrast, writers in the social capital tradition, such as Putnam (1995) and Helliwell and Putnam (1995), emphasize participation. Voluntary activity generates social capital only if it involves personal contact, as in a book club. Giving without social contact, for example, donating to Amnesty International, does not promote what Putnam terms 'civic engagement.' Volunteering involves, in practice, a blend of both participating and giving. It involves participation because volunteering is defined as an activity taking place through an organization or group. It involves giving because it is an unpaid gift of time and effort. The giving and participation dimensions of volunteering create fundamentally different pathways towards social cohesion. GIVING

Gift giving is a ritual found in almost every society. Gifts in premarket economies function as a method of exchange and as a primitive form of insurance because they are given with an expectation that the gift will be

Social Cohesion and Voluntary Activity 161 reciprocated in the future, providing insurance against bad times to come. However, modern large cities are characterized by anonymity and interaction with strangers - hence, gift giving motivated by expectations of future reciprocity is at risk. Nevertheless, social connections still motivate generosity: 'Such information as exists, however, confirms ... ethnic and religious ties, class and organizational loyalties, and nationalism as generators of altruistic actions' (Rose-Ackerman 1996: 714). Gift giving also creates personal ties. As David Cheal (1988: 16) argues, gifts are used 'in the ritual construction of small social worlds.' Gifts build trust by demonstrating intimacy, concern, and a commitment to fulfil obligations. But what about impersonal gifts, such as blood donation? Titmuss argues that 'social gifts and actions carrying no explicit or implicit individual right to a return gift or action are forms of "creative altruism" ... They are creative in the sense that the self is realized with the help of anonymous others; they allow the biological need to help to express itself (1971: 212). Titmuss suggests that there is an intrinsic, even biological, need to help others. This suggestion is consistent with the hypothesis that widely held social norms of altruism, helping or caring for others, can increase a society's probability of survival. As Dayton-Johnson and Friesen suggest (in this volume), the tendency towards altruism may therefore evolve over time. Titmuss also suggests that giving can be a transformative experience, changing our sense of who we are. If a person gives often enough, perhaps she comes to identify herself as, and indeed becomes, an altruistic, caring person. The idea that giving through volunteering causes people to change in fundamental ways is problematic for economists, who generally assume that people's tastes are stable and unchanging. But economists and policy makers have to make a choice. If volunteering is to be explained as rational behaviour, then it is difficult to explain how helping others has benefits outside of homogeneous or exclusive communities small enough to be bound by personal connection. At the same time, explanations of why volunteering might lead to a more giving, caring, inclusive society, as advanced by Titmuss (1971), rest on notions of intrinsic altruism or self-realization which some might find optimistic. PARTICIPATION

Putnam (1995: 665) argues that 'the more we connect with other people, the more we trust them, and vice versa.' Here I suggest four ways way that participation, and connecting with others, creates trust: information, monitoring, group insurance, and sanctioning.

162 Frances Woolley Improved information is the most basic mechanism through which connection can build trust. Meeting others increases knowledge of their character. Because participation provides information about other people's attributes, good or bad, it increases the reliance everyone can place on others, without necessarily increasing altruism, concern, or affective ties between people. Monitoring is an idea that can be traced back to the work of Max Weber (1946). Weber argued that people participating in clubs are trusted - both by club members and by others - because their character is vetted by the club. Admission to an elite association is proof that 'I am a gentleman patented after investigation and probation and guaranteed by my membership' (1946: 308). The same mechanisms have been found in modern associations, such as group lending programs. With group lending programs, group members guarantee each other's loans. They are successful in reducing default rates, in part, because peer monitoring ensures that group members do their utmost to repay (Stiglitz 1990; Wydick 1999). The amount of effort a person is making to repay the loan can be observed by other group members more readily than by, say, a bank or government loan officer (Dayton-Johnson's chapter in this volume discusses peer monitoring in a game theoretic context). A third route from participation to trust is through insurance. Max Weber argued that club members will support each other to maintain the association's reputation. In early Protestant churches, Weber argued, 'In case of need, mutual aid was obligatory' (1946: 318). A key function of the group lending schemes discussed above is insurance. Finally, club members can sanction each other, for example, through social penalties or exclusion from the group. Interestingly, studies of group lending schemes found that sanctions were important, but pre-existing social ties do little to reduce default rates. Indeed, social ties 'may even create a conflict of interest for borrowing groups, making threats of expulsion from the group more difficult and less credible' (Wydick 1999: 474). Is such club behaviour desirable? From the time of Weber to the work of Putnam it has been argued that clubs promote economic growth, because they ensure contracts are kept and loans repaid. But clubs can also have drawbacks. Mafia-type organizations can be thought of as an extremely successful and effective group lending schemes and contract enforcement mechanisms. But an unaccountable quasi-judiciary imposing severe and ruthless sanctions does not create generalized trust. Indeed, mafia-type organizations can be thought of as appropriating the rents to trust in a low-trust society (Friesen, this volume). Group sanctioning may also express the dark side of 'social cohesion,' when a group holds social norms of questionable value. The tight-knit, self-

Social Cohesion and Voluntary Activity

163

reliant communities of Canada's past strictly upheld social norms that would be judged, by today's standards, to be trivial or even dangerous. A woman entering church without her hat on (or a man entering with his hat on) was subject to severe censure, as were other, unacceptable, behaviours (Canadian writers from L.M. Montgomery to Margaret Laurence and Alice Munro give clear pictures of small town life and its self-regulated sanctions). The norm of 'Christian values' has historically led to the exclusion of Jews from many organizations. When monitoring ensures that people do not default on loans, it is perhaps socially desirable. When it means that those not complying with prevailing norms of religion, language, or ethnicity do not receive loans in the first place, it conflicts with modern norms of just social relations. OTHER CONSEQUENCES OF PARTICIPATION

Another salient feature of participatory, rather than giving, organizations, is their tendency towards homogeneity. Each group member will find that the benefits of membership are most closely aligned with the costs when all of the members have similar preferences (Mueller 1989). For example, participants in group lending schemes tend to sort themselves into homogeneous groups (Wydick 1999). The key question then is: Does participation create enclaves of cohesion in a divided society? Or does participation build bridges, reaching those at risk of social exclusion? Whether clubs create enclaves or bridges will depend, first, on, what dimensions matter for club membership. For example, in a chess club, love of chess and chess-playing ability may be the only dimensions along which homogeneous membership is necessary. In a golf club having an income sufficient to pay green fees will matter, and clubs will be more likely to be homogeneous in terms of socioeconomic status. Second, the correlation of dimensions is important. Love of chess may be uncorrelated with income, and hence, chess clubs may create bridges across economic classes. However, if love of chess is correlated with gender, chess clubs may create a division between men and women. A strong system of clubs with overlapping memberships may create a cosmopolitan society as described by Goodin (1996: 364): 'We could be members of many different clubs, drawing on them and contributing to them in turn for many different purposes and many different kinds of support and assistance.' Children's soccer leagues may draw together people from diverse backgrounds and create new social ties. However, clubs with exclusive membership can reinforce social stratification. Games between Glasgow Celtic (Catholic supporters) and Glasgow Rangers (Protestant supporters) can, for example, provide an avenue for the expression of religious conflicts and

164 Frances Woolley reinforce divisions between Protestants and Catholics, while France's 1998 World Cup victory (won by a multiracial soccer team) helped to break down subnational social barriers. Participation and giving have fundamental social effects, creating - and being created by - networks of trust and reciprocity. It remains far from obvious, however, that the benefits from voluntarism spill over to those excluded from social networks. The threat of exclusion may be one way in which social networks build internal trust and enforce reciprocity, even though there is a long tradition, represented here by Titmuss (1971), that argues for the transformative value of giving to others. Summary: The Cohesiveness-Voluntarism Curve The reasons for voluntary work vary from altruism, to commitment, reciprocity, or self-interest. As might be expected, the relationship of voluntary activity with social cohesion depends on its motivation. Voluntary activity may indicate weakness in other social institutions, such as government or family, as well as strength in the voluntary sector. To the extent voluntary activity is linked to cohesion, it is often cohesion among a homogeneous group of the population. Weaving these different theories together, one can hypothesize that the relationship between social cohesion and voluntarism may well have an inverted u-shape. Voluntary activity level may be low in both low-cohesion and high-cohesion societies. Figure 3 illustrates this relationship. On the left-hand side of the diagram would be countries with such a paucity of common values that no voluntary activity is possible. On the right-hand side would be countries with very strong family, church, or other institutions, and possibly also those with great equality of incomes where voluntary activity is unnecessary. In the middle would be countries, such as the United States, where income inequality gives rise to both wealthy people with money to give and people who are in need of charity, where market-induced mobility pulls people away from extended family networks, and where there is no universal church - but where there is a sufficiently strong social fabric that people can turn to voluntary associations and try to do something about poverty and form community networks that compensate for lack of family. Both social cohesion and voluntary activity have many dimensions, and there is no single established index for measuring a society's cohesiveness, nor its voluntarism. Moreover, any comparison across societies would require finding comparable international data. Finally, countries at both ends of the

Social Cohesion and Voluntary Activity 165 Figure 3 The cohesiveness-voluntarism curve Voluntary activity

cohesion scale may well have poor data. Nevertheless, using Canadian data, we can explore the connection between social cohesion and voluntary activity and learn more about the ideas developed so far. Social Cohesion and Voluntary Activity: Canadian Evidence This section uses Canadian data to ask:. • • • •

Is voluntary activity best described as participating or as giving? Does it create enclaves or does it build bridges? What explains the wide variation in voluntary activity across provinces? Does voluntary activity increase Canada's social cohesiveness?

There are three major sources of information about voluntary activity in Canada: the 1987 Survey of Volunteer Activity (VAS), the 1997 National Survey of Giving, Volunteering and Participating (NSGVP), and the 1991 World Values Survey (WVS).

166 Frances Woolley Participating or Giving? Volunteering, as already discussed, involves both giving and participating. The giving element comes from giving time and effort to others; the participation element arises because volunteering means helping others in a particular institutional context, that is, through a voluntary organization or group. The impacts on social cohesion of participating and giving are, however, quite different. The simplest way of finding out why people volunteer is to ask them. According to the WVS, the four most important reasons for volunteering, all ranked as important or very important by approximately 70 per cent of respondents, were 'compassion for those in need,' 'a sense of duty, moral obligation,' 'an opportunity to repay something, give something back,' and 'to make a contribution to my local community.' Of less significance, but still ranked as important or very important by over half of respondents, were 'to help give disadvantaged people hope and dignity,' 'to gain new skills and useful experience,' and 'identifying with people who were suffering.' The WVS, therefore, suggests that altruism is a key motivation for volunteering. Reciprocity (to repay something), local community ties, and gaining work experience are also important.6 The NSGVP, by way of contrast, paints a picture of volunteers with much more direct, personal motivations. The three most commonly cited reasons for volunteering were a personal belief in the cause supported by the organization (cited by 96 per cent of volunteers), to use skills and experience (78 per cent), and having been personally affected or knowing someone who has been personally affected by the cause that the organization supports (over two-thirds). However, because the NSGVP did not ask about broader altruistic motivations, it is impossible to assess their importance from the data. When the NSGVP did allow people to choose 'feel compassion towards people in need' as a motivation, as it did when asking people their reasons for making financial donations, altruism emerged as the single most important motivating factor (all figures from Statistics Canada 1998). In summary, people's motivation for giving is often altruistic, although altruism is bounded by community and personal loyalties. If we shift focus, however, away from why people say they volunteer and, instead, consider what people do when they volunteer, it is clear that volunteer activity involves extensive participation. The three most common voluntary activities involve personal contact: organizing or supervising events (engaged in by over 50 per cent of volunteers), canvassing, campaigning,

Social Cohesion and Voluntary Activity 167 fundraising (45 per cent), and sitting as a board member (38 per cent). Although volunteering does seem to be motivated by concern for others, and a desire to give, the activities engaged in by volunteers do, for the most part, appear to involve participation, predominantly participation within the group or local community. This raises a recurring question: Does voluntary activity create enclaves or build bridges? Enclaves or Bridges? Does voluntary activity create clubs of homogeneous members that may be elite enclaves of trust and reciprocity? Or is voluntary activity a way by which people reach out to others, a bridge that unites Canadians of different •races, classes, ages, and backgrounds? This section examines what people do when they volunteer, the types of organizations they volunteer for, and the characteristics of volunteers. It asks: Do people at risk of social exclusion engage in voluntary activity? Some volunteer activity is centred around outreach to others. NSGVP data indicate that over 20 per cent of Canadians provide care or support as a volunteer, and 15 per cent collect, serve, or deliver food to others. However, these reaching-out activities are among the less common forms of volunteer work. The most common forms of volunteer work - organizing events and sitting on boards - are likely to involve personal contact primarily with those connected with the association and, thus, are in some sense exclusive. Even canvassing, campaigning, and fundraising may well be locally centred (e.g., selling Girl Guide cookies), and the excluded are not likely to be involved. Because the average level of donations rises with income, higher-income Canadians are a better target for fundraising than those with lower incomes. Analysis of the organizations that volunteers work for (using NSGVP data) confirms this same basic pattern. Social service organizations (this category comprises the bulk of organizations that work with those at risk of social exclusion) account for 21 per cent of volunteer hours. Social clubs and sports and recreational organizations, which tend to provide services primarily to their members, account for 28 per cent of volunteer hours. Religious organizations, which again serve a defined community, account for 18 per cent of hours. The remaining hours are accounted for by health organizations (10 per cent) and 'other' (23 per cent). Again, the pattern emerges that a small proportion of volunteer activity overall involves reaching out to those outside the immediate community, while the greater number of volunteer hours are centred around local groups of which the volunteers are members.

168 Frances Woolley Analysis at the level of the individual also casts doubt on the idea that voluntarism builds bridges. People with more interactions and more connections with others are more likely to volunteer. In the NSGVP data, volunteering rates rise steadily with household income, from 22 per cent when household income is below $20,000 to 44 per cent when household income is $80,000 or more. The pattern is even more pronounced with education: 21 per cent of people with less than a high school education volunteer, compared with 48 per cent of those with a university degree. Volunteer rates rise with age, peaking at ages 35 to 44, when 37 per cent of people volunteer, and declining to a rate of 23 per cent for those age 65 and over. Regression analysis with other Canadian data confirms the same basic patterns. Day and Devlin (1996) found that women, people of both sexes with more education, people with strong religious beliefs, and people with good health were more likely to volunteer. People who speak a language other than English or French at home were less likely to volunteer. Vaillancourt (1994) arrived at similar conclusions, and (as already noted) he also found a positive relationship between volunteering and Protestantism. All this suggests that there are positive correlations along dimensions of belonging, participating, and inclusion. The reason for this is simple: People volunteer because they are asked to do so. 'Volunteers explain that they don't do more, and non-volunteers explain that they don't volunteer at all, due to the fact that no one ever asked them' (Statistics Canada 1998: 41). Voluntarism is the product of the embeddedness of individuals in relationships. Family, community, and employment connections strengthen volunteering, but volunteering also strengthens labour market and other skills. Over threequarters of the volunteers surveyed in the NSGVP reported that as a result of volunteering they gained interpersonal skills such as understanding people better, learning to motivate others, and learning to cope with difficult situations (Statistics Canada 1998: 40). The World Values Survey further supports the pattern of the connected volunteer and the more isolated non-volunteer. People who are 'on top of the world / feeling that life is wonderful' volunteer more often (47 per cent; standard error =1.6 per cent) instead of 37 per cent (standard error [s.e.] = 1.7 per cent). Conversely, people who feel 'very lonely or remote from other people' volunteer less often, as do people who feel 'bored' - 39 per cent of the lonely volunteer (s.e. = 2.8 per cent), as do 38 per cent of the bored (s.e. = 2.2 per cent), compared with 44 per cent of those who are not lonely (s.e. = 1.3 per cent) or bored (s.e. = 1.4 per cent) (numbers are calculated from the World Values Survey 1991). The interrelationships here may not be so strong as to be described as 'cohesive enclaves.' They do suggest, however, that voluntary activity, em-

Social Cohesion and Voluntary Activity

169

ployment, family, and community create a virtuous cycle of involvement, while loneliness, unemployment, and poverty are also mutually reinforcing. Interprovincial Variations in Voluntary Activity If volunteering has psychological and economic benefits, then Canadian policy makers have an interest in finding out why volunteerism rates differ so widely across provinces. Table 1 shows different measures of voluntary activity by province. Quebec has lower levels of voluntary activity than the Canadian average and - with one or two exceptions - lower levels of voluntary activity than are found in any other province. Saskatchewan stands out as being a province of volunteers. These interprovincial variations represent a challenge to the idea that voluntary activity is in some way linked to social cohesion. Quebec is arguably one of the most cohesive provinces in Canada. Quebec social and economic policy reflects a public commitment to reducing disparities in income and wealth. Quebec has a distinctive linguistic and cultural heritage, providing a shared community of interpretation, at least for the majority Francophone population. Vaillancourt (1994) may be overstating the case somewhat when he writes, 'On the basis of language, ethnicity, or religion, Quebec is a much more homogeneous region of Canada than the other four' (1994: 818). However, compared with the other large provinces (British Columbia, Alberta, and Ontario) Quebec does have much lower levels of immigration (less than 10 per cent of the Quebec population are immigrants, compared with about 20 per cent for the other large provinces) and a smaller visible minority population.7 To the extent that recent immigrants and members of visible minorities are likely to have different values and cultures from other Canadians, Quebec's smaller immigrant and minority population would be expected to enhance the sense of shared values. Quebec is also remarkably homogeneous with respect to religion. According to the 1991 census, 86 per cent of the Quebec population is Roman Catholic. The only other provinces that come close to this level of religious homogeneity are Newfoundland (with 61 per cent Protestant) and Saskatchewan (with 54 per cent Protestant), and this ignores the divergence among Protestant religions. What can explain the low level of voluntary activity in Quebec? Vaillancourt (1994) has argued that perhaps we would expect to find less volunteer activity in homogeneous regions because 'voters in more homogeneous regions are more likely to agree on what goods and services they wish the public sector to provide and thus need less volunteer work' (1994: 818). Also, Quebec society may be less homogeneous than it appears. Although the ma-

170 Frances Woolley Table 1 Voluntary activity by province

Newfoundland Prince Edward Island Nova Scotia New Brunswick Quebec Ontario Manitoba Saskatchewan Alberta British Columbia

Donor rate (%)

Volunteer rate (%)

Membership rate (%)

84 83 83 82 75 80 81 83 75 73

33 36 38 34 22 32 40 47 40 32

49 50 55 47 43 52 58 62 55 54

Source: NSGVP, Statistics Canada (1998)

jority of Quebecois share a common religious heritage and ethnicity, Quebec society is divided along other lines: separatist/federalist, class, region, rural/ urban, and political affiliation. Perhaps homogeneity on one or two dimensions does not guarantee social cohesion. A third possibility is that the interprovincial variations in voluntary activity reflect religious differences across the provinces. The high-voluntarism provinces, Saskatchewan and Newfoundland, have a majority of Protestants, whereas the low-voluntarism province, Quebec, is strongly Catholic. As argued earlier, Protestantism appears to be associated with social norms supporting voluntarism. A simple analysis of the interprovincial differences finds a strong positive relationship between the percentage of the population that is Protestant and measures of voluntary activity: the correlation coefficient r is 0.64 between Protestantism and membership of associations, 0.74 between Protestantism and volunteering, and 0.82 between Protestantism and the median amount donated. Evidently, the level of voluntary activity may not be strongly correlated with shared values or a common cultural heritage. Furthermore, if voluntarism is a Protestant phenomenon, great caution needs to be exercised in promoting a voluntarist ethic in societies with other religious traditions. Any relationship that we find between social cohesion indicators and voluntary activity may not necessarily cross cultural lines. Or, Quebec data on voluntary activity may simply mean that in this, as in other things, Quebec is a 'distinct society.'

Social Cohesion and Voluntary Activity 171 Does Voluntary Activity Build Social Cohesiveness? 'Good fences make good neighbors' Mending Wall, Robert Frost (1963)

Are volunteers different? And, if so, how? This section will examine measures of social cohesiveness, comparing volunteers and non-volunteers, using the World Values Survey. The WVS offers two possible indicators of volunteering: belonging to an organization (belongers) or doing work for an organization (volunteers). This section examines both, since the two at times have different implications for measures of cohesiveness. One well-established difference between volunteers and non-volunteers is that volunteers trust more. Among Canadian respondents to the 1991 World Values Survey, 58 per cent (s.e. = 1.8 per cent) of those who do unpaid work believe that most people can be trusted, compared with 48 per cent of those who do not do unpaid work (s.e. = 1.6 per cent). If we turn from generalized trust, to trust in specific groups of people, there is much less of a difference between volunteers and non-volunteers. Trust in family (98 per cent), own countrymen (80 per cent), and Americans (54 percent) was essentially the same for volunteers and non-volunteers. Volunteers are more trusting of immigrants: 48 per cent of volunteers trust immigrants (s.e. = 1.7 per cent) compared with 43.7 per cent (s.e. = 1.6 per cent) of non-volunteers. Volunteers are slightly more trusting of Mexicans, Russians, and Chinese than are non-volunteers, but the differences are small (less than three percentage points) and statistically insignificant.8 The responses of Canadian volunteers can be contrasted with other groups surveyed by the WVS. Americans and Canadians, both volunteers and nonvolunteers, had similar levels of trust in family and own countrymen. When asked about other groups, for example, immigrants or Mexicans, American volunteers are nine percentage points more trusting than American non-volunteers. However, in Canada there is a much smaller gap between volunteers and non-volunteers. For both groups, 58 per cent trust French Canadians (English Canadians in Quebec) and 54 per cent trust Americans. While there is a gap in trust of immigrants - 48 per cent of volunteers trust immigrants (se 0.018) compared to 43.6 per cent of non-volunteers (se 0.016) - the gap is smaller than in the United States. This suggests that the link between voluntarism and increased trust may be country specific. Belongers are people who belong to an association. Belonging measures membership or participation, in contrast to volunteering, which is about doing unpaid work for an association.

172 Frances Woolley Considering people who belong to associations (participation) there is a pattern of uniformly decreasing trust, as we move from family (trusted completely or trusted by 98 per cent of respondents, both belongers and nonbelongers) to own countrymen (80 per cent) and Americans (54 per cent). The only substantial difference was in attitudes to immigrants, who were trusted by 48 per cent of belongers (s.e.=1.5%) and 41 per cent of nonbelongers (s.e.=2.0%). These results suggest an amendment to Putnam's hypothesis that 'the more we connect with other people, the more we trust them.' All of those surveyed trust people they know, for example, family or own countrymen. However, volunteers and belongers are more likely to trust newcomers or outsiders, specifically immigrants. This supports either a Titmuss-type idea that giving is a transformative experience, leading to generalized concern for others, or a selection effect, that is, the cultural values or socioeconomic background of people who volunteer (see Knack and Keefer 1997). Another measure of social cohesion is tolerance. The World Values Survey measures tolerance by asking people to identify groups that they 'would not like to have as neighbours.' Table 2 shows the percentage of Canadian respondents mentioning each of the named groups whose members they would not like as neighbours. People who do, and those who do not, volunteer have strikingly similar tolerance levels. Their attitudes towards living next to people of a different race, emotionally unstable people, Muslims, Jews, and Hindus are almost identical. Although there is some racism, particularly towards Muslims and Hindus, a large majority of Canadians are tolerant of racial and ethnic differences. The categories mentioned least often were Jews, people of a different race, immigrants or foreign workers, large families, Muslims, and Hindus. Volunteers are, however, significantly less tolerant of extremist or 'antisocial' behaviour, for example, having a criminal record or being a heavy drinker. Members of religious organizations are the least likely to tolerate anti-social behaviour, and they also are less likely to tolerate homosexuals or people with AIDS as neighbours. These differences between volunteers and non-volunteers suggest that people who volunteer are the centrist 'pillars of society' and that they are intolerant of political extremism, those who break society's rules (criminals), and those who deviate from social norms. This supports the argument that the voluntarist ethic is maintained, in part, by imposing sanctions on those who violate social norms. A more positive interpretation is that volunteers care about who their neighbours are because they are involved in neighbourhood activities. Why worry about living next to right-wing extremists if you never talk to your neighbours? Why worry about

Social Cohesion and Voluntary Activity

173

Table 2 Voluntary activity and tolerance Groups identified that 'You would not like to have as neighbours' - percentage of respondents mentioning group, by organizational membership, Canada, 1991

With criminal record

Member of any organization

Do any voluntary work

Belong to religious organization

Nonmember

Nonvolunteer Volunteer

Do not belong

40.3 (1,56) 5.0 (0.70) 23,8 (1.36) S2.7 (1.60) 21.3 (1.31)

39.4 (1.36) 5.2 (0.62) 25.9 (122} 52.3 (1,39) 23.3 (1.17)

40.6 (1.97)

Member

43.4 (1.49) 4.7 (0.64) 29,4 (1.37) 56.1 (1-49) 26.9 (1.37) 4,7 (0.64)

5.1 (0.88) 22,6 Left wing extremists (1.68) 51.6 Heavy drinkers {2.0} Right wing extremists 20.2 (1. 61) Large families 7,9 (1.08) 29.7 Emotionally unstable 29.4 (1.37) (1.83) 10.7 9.8 Muslims (0.93) (1.19) Immigrants/foreign 4.7 7.5 workers (0.64) (1.06) 20.9 20.1 People with AIDS (1.22) (1.61) Drug addicts 64.0 61.1 (1.44) (1.95) 28.7 31.4 Homosexuals (1.86) (1.36) 5.3 5.9 Jews (0.67) (0.95) 10.2 10.1 Hindus (1.21) (0.91) 623 1107 Sample size (n=)

Different race

45.2 (1.82)

72 (0.82) 30.1 (1.46) 9.5 (0.94)

4.7 (0.77) 31,2 (1.70) 56.7 (1,82) 28.6 (1.66) 5.3 (0.82) 28.9 (1.66) 11.5 (1.17)

6.3 (0.77) 20.2 (1.28) 62.5 (1.54) 29.4 (1.45) 5.6 (0.74) 9.5 (0.94) 984

5.0 (0.80) 21.2 (1.50) 63.6 (1.76) 30.1 (1.68) 5.3 (0.82) 10.9 (1.14) 746

Belong

6.7 (0.69) 29.3 (1.26) 9.8 (0.82)

51.3 (2.41) 3.9 (0.94) 30.1 (2.21) 61.0 (2.35) 27.8 (2.16) 5.4 (1.09) 30.3 (2.21) 12.1 (1.57)

6.1 (0.66) 18,9 (1.09} 61.2 (1.35) 27.8 (1.24) 5.7 (0.64) 10.1 (0.83) 1298

4.6 (1.01) 25.9 (2.11) 68.4 (2.29) 35.3 (2.30) 4.9 (1.04) 10.4 (1.47) 432

Source: Author's calculations using the 1991 World Values Survey Notes: The shaded boxes indicate a significant difference with 95% confidence between the non-member/non-volunteer/those not belonging and those who are members/do belong. The hypothesis tested is whether or not the point estimate of the member is significantly different from the non-member. The question asked by the WVS is 'On this list are various groups of people. Could you please sort out any that you would not like to have as neighbours?'

174 Frances Woolley the heavy drinkers next door if you do not open your door when someone knocks? Perhaps, as Robert Frost wrote in his poem Mending Wall (1963), good fences do make good neighbours, establishing boundaries and norms of acceptable behaviour. Yet, one has to observe, as Frost does, Before I built a wall I'd ask to know What I was walling in or walling out

Any fence brings with it a risk of social exclusion. An alternative measure of tolerance is attitudes towards those who live in need. When asked, 'Why are there people in this country who live in need?' people who volunteer are most likely to give, as their first response, 'Because there is injustice in our society.'9 Canadians who do not volunteer are most likely to believe people live in need 'because of laziness and lack of will power.' This could be taken as support for Titmuss's idea that giving increases compassion for others. Alternatively, caring people may be more likely to volunteer, or perhaps the relationship is driven by a third factor, such as religious conviction. One final measure of social cohesion is the values that people attempt to instil in their children. The World Values Survey asked respondents to choose five values or behaviours that they consider important for children to learn. Canadians chose tolerance and respect for others more than any other value. Volunteering made little difference to the value placed on tolerance. Tolerance and respect were chosen by exactly the same proportion - 80 per cent of those who did and those did not do unpaid work, as well as by 81 per cent of belongers and 78 per cent of non-belongers. The greatest difference between volunteers and non-volunteers was in the importance placed on good manners. 'Good manners' was rated as much more important by non-volunteers: it was chosen by 78 per cent of non-volunteers and 81 per cent of nonbelongers, as opposed to just 71 per cent of those doing unpaid work for or belonging to organizations - a statistically significant difference. Religious faith was rated as much more important by volunteers, chosen by 35 (33) per cent of volunteers (belongers), as compared with 27 per cent of non-volunteers (again statistically significant at the 95 per cent level). This is not surprising, given the strong correlation between religious conviction and volunteering. Apart from this, there are no major differences. Volunteers place a slightly higher priority on independence, hard work, responsibility, imagination, and determination, while non-volunteers place a slightly higher priority on thrift, unselfishness, and obedience. All the differences are small. All the differences, except for thrift and responsibility, are statistically insignificant.

Social Cohesion and Voluntary Activity 175 Clearly, volunteers and non-volunteers do differ in their attitudes and values. Volunteers have greater trust in strangers and less tolerance of deviance. They are more likely to believe injustice exists. They are more concerned with teaching children religious conviction and less concerned about good manners. However, there is no compelling reason to call this particular constellation of values social cohesiveness. On many basic measures, such as attitudes towards having people of a different race as neighbours or teaching children tolerance and respect, there is a broad consensus among all Canadians that tolerance is important. Volunteering appears to have little impact on this consensus. The Potential, and Limits, of Voluntary Action Although the relationship between social cohesion and voluntarism may not be straightforward, does voluntarism have advantages? Is there any evidence that the voluntary sector provides services better than government or the private sector? If so, which ones? What are the major advantages and disadvantages of voluntary provision? The arguments presented here will suggest that the voluntary sector is best at providing a fairly small quantity of highquality, personal goods and services. Policy makers may also be interested to know how the amount of voluntary activity could be increased. For any service, there is a limit on the quantity that can be provided voluntarily. However, providing payment to motivate greater supply may be dysfunctional. A number of authors have hypothesized that switching from voluntary to paid provision will initially decrease provision (see, e.g., Frey 1993). With the introduction of a market, people can calculate the obvious monetary costs and benefits of their actions, and these obvious cost and benefits may shift attention away from, crowd out, or diminish the intangible value of giving. A second case for voluntary provision is that voluntarily provided services are of higher quality. Perhaps volunteers are more caring, providing more personal service? Perhaps paying for service changes the nature of the service provided? As Julie Nelson asks: 'Would higher pay somehow drain away the emotional and interpersonal elements that we suppose to be in [caring] work currently and that make a recipient feel truly "cared for"? Or might higher pay create an influx of pseudo-carers who gravitate to such jobs "just for the money"?' (1999:44). Could monetary motivations overwhelm or crowd out intrinsic motivation, resulting in dysfunctional optimizing? One way of sorting out these ideas is by returning to Figure 1. One issue is how much paid caring work (work in the lower half of the diagram) should

176 Frances Woolley be paid. Payment appears to affect motivation, but in subtle ways. Frey finds that 'External interventions [such as pay] crowd out intrinsic motivation if they are perceived to be controlling and they crowd in intrinsic motivation if they are perceived to be acknowledging' (1996: 11, emphasis in original). For paid work, acknowledging pay can crowd in intrinsic motivation and at the same time improve the status of women, who make up the bulk of the caring sector workforce. This makes a case for paying caregivers well. A separate issue is how caring work should be organized: through markets, through governments, through families, or through volunteer and non-profit organizations? It is consistent to argue, for example, that child care workers should be well paid and, at the same time, promote non-market forms of provision. But then this raises the question, What are the advantages of nonmarket provision? In the service sector, it is common, particularly in the United States, to have non-profit and for-profit nursing homes, day care centres, and hospitals operating side by side. Rose-Ackerman (1996: 722-3) finds that non-profit nursing homes and day care centres occupy a different market niche from for profit-providers, offering higher-quality services, often at a higher cost. Nonprofit nursing homes receive fewer complaints, and non-profit day care centres rank more highly on measures of input quality such as child-staff ratios and teacher training. Studies of hospitals, however, have found that organizational form is not associated with differences in quality or cost, once differences in patient mix are taken into account. This finding is in line with the literature on privatization, which has found that market structure matters more than ownership (public or private). When government-owned industries and privately owned industries have operated side by side in a competitive environment, for example, Canadian National and Canadian Pacific, Air Canada and Canadian, they produce goods and services equally efficiently (Cullis and Jones 1998). In general, when non-profit and for-profit firms operate in the same environment, producing the same services and competing for the same customers, they will tend to produce similar quality goods and services. If for-profit companies can provide comparable goods and services to nonprofits, why do non-profit providers emerge? Rose-Ackerman (1996: 719) suggests ideology is a key reason for non-profit entry. The ideologue, for example, someone with a particular religious conviction or approach to caregiving, will be attracted to the non-profit form because of the lack of owner-investors, who would question ideologically motivated investments. Rose-Ackerman's work hints at a conflict between the non-profit form and accountability. Because volunteers provide labour voluntarily - because they

Social Cohesion and Voluntary Activity 177 want to - it is fundamentally difficult to make them 'accountable' to the job demands of others. Voluntary service provision can conflict with responsiveness to client needs. Frey (1993: 646-7), surveying psychological studies on intrinsic motivation, concludes that the more rewards are 'contingent on task engagement and on the performance desired' the more intrinsic motivation will be crowded out. Volunteers who are required to provide services according to someone else's specifications lose their intrinsic motivation. Indeed, when volunteers are required to work to particular specifications, they become little more than workers who are just like any others, except that they are not paid. Voluntary provision implies complex power relationships. If service providers are required to perform tasks to a certain standard, power is in the hands of managers or service recipients. However, if service providers' only reward is job satisfaction, they will need to derive intrinsic motivation from, for example, becoming actively involved in decision making or having control over their work. Decision-making ability and control will place power in their own hands. Gifts are sometimes seen by anthropologists as 'first and foremost a means of controlling others' (Mauss, cited in Cheal 1988: 3). In the same way, volunteers whose motivation is job satisfaction have social power over service recipients and the realistic option of ignoring managerial directives. Private charity may also be less universal, and more discriminating, than government assistance. People are more likely to be intrinsically motivated when they have 'extensive ... participation possibilities' (Frey 1993), that is, they are involved in decision making, such as determining who gets service. This may conflict with government-determined criteria for benefit eligibility. Moreover, intrinsic motivation appears to be stronger when there is a personal relationship between people. Voluntary agencies will, therefore, tend to be most successful in helping people who have personal ties to the agency. In one sense, this can be extremely positive, if volunteers build personal ties and become engaged with the people they are helping It is less unambiguously positive if it means that voluntary workers help most those with whom they have personal ties of language, religion, or ethnicity. Universalistic, nondiscriminatory norms are central to the modern liberal state, but not to the motivation of voluntarism. If service delivery is, for example, devolved to religious organizations, what happens to the non-religious, those without a simple community identification? If governments wish to replace the public provision of services with other forms of delivery, the discussion here suggests that universal access to services and empowerment of service recipients are key areas of concern.

178 Frances Woolley Conclusions Will a reduction in public sector functions create the space for voluntary activity that brings with it stronger social cohesion? The analysis in this essay suggests that such an outcome is unlikely. There is little evidence that reductions in government spending lead to greater voluntary activity. Rather, shared values, the absence of extreme income inequality, and shared tastes are necessary for collective action - whether through governments or voluntary organizations. Income inequality and social fragmentation erode the basis for voluntarism, just as they put stress on 1960s and 1970s models of government. Other findings suggest the Northern European model of voluntary activity will only be adopted slowly elsewhere. Voluntarism is strongly related to particular social norms and institutions, for example, non-hierarchical religions and small nuclear families. This calls into question the possibility, much less the desirability, of imposing the voluntary model on those with other religious traditions and family networks. Even if the voluntary sector can, to some extent, offset reductions in government, it is far from obvious that this will result in greater social cohesion. It is true that there is some evidence linking voluntary activity to trust. However, there appears to be no relationship between voluntarism and other social cohesion indicators, such as shared values or tolerance. If anything, voluntarism appears to be associated with less tolerance of social deviance. Moreover, only a small proportion of voluntary activity appears to create bridging social networks, reaching those at risk of social exclusion. Instead, employment, marriage, having children, and more education appear to create a virtuous cycle where voluntarism reinforces, and is reinforced by, existing social networks. Turning to evidence on the comparative performance of non-profit and voluntary versus for-profit and government organizations, it appears that the key difference between non-profit and other institutions is in terms of motivation. Non-profits are more likely to be motivated by ideology. Volunteers rely on intrinsic motivation, which is contingent upon personal involvement and control over tasks, suggesting voluntary services will tend to be more discriminating and less universal than government-provided services. Finally, there is overwhelming evidence that market structure matters more than organizational form. Given the difficulty in arriving at a clear interpretation of social cohesion, and the lack of understanding of factors that promote or hinder it, economists and governments would be well advised to shift focus to social exclusion.

Social Cohesion and Voluntary Activity 179 Social exclusion, being an individual-level concept, is more amenable to economic analysis, which also operates largely at the individual level. Social exclusion also focuses attention on individuals at risk, and at urgent problems, such as unemployment, poverty, isolation, and discrimination. Notes 1 I would like to thank Paul Bernard, David Long, Jane Jenson, Paul LeDuc Brown, Shelley Phipps, Lars Osberg, Jeff Dayton-Johnson and Dick Stanley for helpful comments and conversations, Paul Conroy, Ambrose Leung and Lynn Lethbridge for very capable research assistance, and Heritage Canada for financial support. 2 Cited in Policy Research Initiative (1999) Horizons vol 2(2), p. 4. 3 Except for income data for Luxembourg (1985), Germany (1984), and France (1984). In all other cases income data is for 1990, 1991, or 1992. 4 The Anglican church is exceptional in being a Protestant church with an organizational structure which closely parallels the Roman Catholic church. 5 Statistics Canada web site, 25 Nov. 1999; www.statcan.ca/english/Pgdb/People/ Families/famil36a.htm. 6 Figures calculated by the author from World Values Survey, Canada (1991). 7 The visible minority population is smaller than that of the other large provinces around 6 per cent, compared to 10 per cent in Alberta, 15 per cent in Ontario, and 18 per cent in British Columbia. 8 Mexicans are trusted by 38.3 per cent of volunteers (1.8%) versus 35.8 per cent of non-volunteers (s.e. = 1.6%); Russians are trusted by 39.0 per cent (s.e. 1.8%) and volunteers and 37.0 per cent (s.e. 1.6%) of non-volunteers; Chinese are trusted by 43.0 per cent of volunteers (s.e. 1.9%) and 42.0 per cent (s.e. 1.6%) of non-volunteers. 9 Injustice: 35.8 per cent of volunteers (s.e. = 1.8%); 28.8 per cent of non-volunteers (s.e. = 1.5%). Laziness: 35.2 per cent of non-volunteers (s.e. = 1.5%), 27.2 per cent of volunteers (s.e. = 1.7%), figures calculated from World Values Survey, Canada (1991).

References Allen, R.E. 1990. The Concise Oxford Dictionary. Oxford: Clarendon Press. Andreoni, James. 1995. 'Cooperation in Public-Goods Experiments: Kindness or Confusion.' American Economic Review 85(4): 891-904. - 1988. 'Privately Provided Public Goods in a Large Economy: The Limits of Altruism.' Journal of Public Economics 34: 57-73.

180 Frances Woolley Cheal, David. 1988. The Gift Economy. London and New York: Routledge. Cullis, John, and Philip Jones. 1998. Public Finance and Public Choice, 2nd ed. Oxford: Oxford University Press. Day, Kathleen, and Rose Anne Devlin. 1996. 'Volunteerism and Crowding Out: Canadian Econometric Evidence.' Canadian Journal of Economics 29(1): 37-53. - 1998. 'The Payoff to Work without Pay: Volunteer Work as an Investment in Human Capital.' Canadian Journal of Economics 31(5): 1179-91. Ferris, J. Stephen, and Edwin G. West. 1998. 'Private versus Collectivized Charity: Further Explorations of the Crowding Out Debate.' Photocopy, Department of Economics, Ottawa, Carleton University. Frey, Bruno. 1993. 'Motivation as a Limit to Pricing.' Journal of Economic Psychology 14: 635-64. - 1996. 'Institutions and Morale: The Crowding-Out Effect.' Paper prepared for the Conference on the Relationship between Economic Institutions and Ethical Values and Behavior, Yale University, New Haven, Conn., April. Frost, Robert. 1963. Selected Poems of Robert Frost. New York: Holt, Rinehart and Winston. Goodin, R. 1996. 'Inclusion and Exclusion.' Archives Europeenes de Sociologie 37: 343-71. Gottschalk, Peter, and Timothy M. Smeeding. 1997. 'Cross-National Comparisons of Earnings and Income Inequality.' Journal of Economic Literature 35(2): 633-87. Helliwell, John. 1998. 'Do Borders Matter for Social Capital? Economic Growth and Civic Culture in U.S. States and Canadian Provinces.' Paper presented at the conference Policy Research: Creating Linkages, Ottawa, 1-2 Oct. Helliwell, John F., and Robert D. Putnam. 1995. 'Economic Growth and Social Capital in Italy.' Eastern Economic Journal 21(3): 295-307. lannaccone, Laurence R. 1998. 'Introduction to the Economics of Religion.' Journal of Economic Literature 36(3): 1465-95. Jenson, Jane. 1998. 'Mapping Social Cohesion.' Plenary Address given at the conference Policy Research: Creating Linkages, Ottawa, 1-2 Oct. - 1999. Mapping Social Cohesion: The State of Canadian Research CPRN Study, No. F03. Ottawa: Canadian Policy Research Networks. Knack, Stephen, and Philip Keefer. 1997. 'Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.' Quarterly Journal of Economics 112(4): 1251-88. Luxton, Meg. 1998. 'Families and the Labour Market: Coping Strategies from a Sociological Perspective.' In David Cheal, Frances Woolley, and Meg Luxton, How Families Cope and Why Policymakers Need to Know. Ottawa: Renouf Publishing.

Social Cohesion and Voluntary Activity

181

Mueller, Dennis C. 1989. Public Choice II. Cambridge, Mass.: Cambridge University Press. Nelson, Julie. 1999. 'Of Markets and Martyrs.' Feminist Economics 5(3): 43-60. Organization for Economic Co-operation and Development (OECD). 1997. Societal Cohesion and the Globalising Economy: What Does the Future Hold? Paris: OECD. Ostrom, Elinor, and James Walker. 1997 'Neither Markets nor States: Linking Transformation Processes in Collective Action Arenas.' In Dennis C. Mueller, ed., Perspectives on Public Choice: A Handbook. Cambridge: Cambridge University Press. Pearsall, Judy, ed. 2001. The Concise Oxford Dictionary, 10th ed. rev. Oxford: Oxford University Press. Phipps, Shelley. 1998. 'Social Cohesion and the Well-Being of Canadian Children.' Paper prepared for the conference Policy Research: Creating Linkages, Ottawa, 1-2 Oct. Putnam, Robert D. 1993. Making Democracy Work; Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. - 1995. 'Tuning In, Tuning Out: The Strange Disappearance of Social Capital in America.' PS Political Science and Politics (Dec.): 664-83. Rose-Ackerman, Susan. 1996. 'Altruism, Nonprofits, and Economic Theory.' Journal of Economic Literature 34(2): 701-28. Statistics Canada. 1998. Caring Canadians, Involved Canadians: Highlights from the 1997 National Survey of Giving, Volunteering and Participating. Ottawa: Statistics Canada. Stiglitz, J. 1990. 'Peer Monitoring and Credit Markets.' World Bank Economic Review 4: 351-66. Strategic Research and Analysis Directorate. 1977. 'Canadian Identity, Culture and Values: Building a Cohesive Society.' Ottawa: Heritage Canada. Sugden, Robert. 1984. 'Reciprocity: The Supply of Public Goods through Voluntary Contributions.' Economic Journal 94: 772-87. Titmuss, Richard. 1971. The Gift Relationship. London: George Allyn and Unwin. Todd, Emmanuel. 1985. The Explanation of Ideology: Family Structures and Social Systems. Oxford: Basil Blackwell. Vaillancourt, Francois. 1994. 'To Volunteer or Not: Canada 1987.' Canadian Journal of Economics 27(4): 813-26. Weber, Max. 1946. The Protestant Sects and the Spirit of Capitalism.' In H.H. Gerth and C. Wright Mills, eds., From Max Weber: Essays in Sociology. New York: Oxford University Press.

182

Frances Woolley

Woolley, Frances. 1998. 'Work and Household Transactions: An Economist's View.' In David Cheal, Frances Woolley, and Meg Luxton, How Families Cope and Why Policymakers Need to Know. Ottawa: Renouf Publishing. Wydick, Bruce. 1999. 'Can Social Cohesion be Harnessed to Repair Market Failures? Evidence from Group Lending in Guatemala.' Economic Journal 109 (July): 46375.

Communities and Economic Prosperity; Exploring the Links JANE FRIESEN1

Questions of economic policy are often framed in a way that market efficiency and social equity are seen as contradictory goals. Recently, however, a more nuanced understanding of the relationship between social goals and economic efficiency has begun to evolve, as social scientists from a variety of disciplines have turned their attention to the important role that social relationships may play in fostering economic prosperity. Social interactions that do not have an explicit economic purpose may affect economic productivity in a variety of ways. In this chapter, I will focus on social interactions that take place in a community. By a community I mean a group of people who experience repeated, direct, face-to-face social interactions because of a characteristic they have in common. A community might be defined, for example, by a geographic boundary surrounding a neighbourhood school, by an ethnic or religious group whose members participate in the same institutions, or by a group of individuals who share a workplace. Face-to-face social contact produces and supports basic social norms that are economically important. Language norms, for example, enable the communication that is fundamental to economic coordination. Community social norms (e.g., punctuality or honesty), established through ongoing social contact, may increase the level of trust among members of a group, reducing transaction costs, deterring opportunistic behaviour, and internalizing externalities. Supportive social relations may improve psychological and emotional health, which in turn affect individual productivity. Knowledge can be transferred in a variety of community and social settings. Community institutions and information channels and networks are opened through community social contact.

184 Jane Friesen Social interactions can also have less benign effects. Peer groups may influence members to engage in harmful behaviours such as drug abuse, vandalism, and teen-age childbearing. Racism and sexism may undermine individuals' self-esteem and cut them off from important information channels and networks, and social networks may reinforce racist and sexist norms. In each of these examples, social contact generates some form of economic spillover across individuals because the behaviour of one individual has economic as well as social consequences for the individuals with whom they interact socially. Questions of community social organization, therefore, may have important economic implications because the way communities are organized will structure the quantity and quality of social contact. The first section begins by reviewing some of the literature in economics that bears on the relationship between community relations and economic outcomes. As well as identifying ways in which social contact among members of a community is economically important, economic theory can provide important insights into the nature of the institutional and organizational structures that will support economically productive social relationships. The second section focuses on education policy. The third section presents the results of an empirical investigation of the relationship between a number of neighbourhood socioeconomic characteristics and school-level outcomes on standardized Grade 12 exams in British Columbia. The fourth section provides concluding remarks. The Economic Importance of Community Social Interaction The ways through which community social interactions affect economic prosperity can be usefully grouped into three broad categories: (1) the formation of social norms, (2) the creation of information networks, and (3) human capital spillovers. Norm formation refers to the way in which a person's own behaviour is conditioned by expectations of the behaviour of others within the community, given a person's information and human capital. The creation of information networks concerns the way in which information about options and choices is transmitted to an individual with a given level of skills, through social networks. Human capital spillovers refers to the way in which the acquisition of skills by an individual is influenced by the decisions and skills of other members of the community. This section provides a review of some of the theoretical and empirical literature in economics in each of these categories.

Communities and Economic Prosperity

185

Social Norms and Values Ben-Ner and Putterman (1998) provide an extensive discussion of the economic importance of values and norms. They define an individual's values somewhat narrowly as process-regarding preferences, which concern 'the manner in which they themselves and others behave, including the ways in which they attain outcomes of interest.'2 I have in mind something more like the broader definition of Schwartz (1993): 'principles, or criteria, for selecting what is good (or better, or best) among objects, actions, ways of life and social and political institutions and structures.'3 Most authors refer to a social norm as 'something "external to actors" or resulting from the interaction of a number of actors.'4 Values and norms may play an important economic role in at least three ways. First, they condition the decisions of individuals when those decisions have an unintended effect on a bystander. Day ton-Johnson's chapter (in this volume) discusses the conditions under which norms of cooperation are stable. Often norms of good behaviour (such as picking up after your dog) serve to reduce negative externalities. Decisions made without taking into account third-party effects or externalities are economically inefficient because the decision maker does not bear the full cost of her action or does not reap the full benefit. The 'invisible hand' of the marketplace does not guide her to make the choice that would be most desirable if all of the costs and benefits experienced by all those affected were taken into account. Altruistic values may lead people to internalize these third-party effects, improving the economic efficiency of the outcome. Second, broadly based social norms may standardize behaviour (an example is social norms of punctuality). This may reduce costly uncertainty in transactions between individuals, thereby increasing social welfare, or it may force diverse individuals into narrowly conformist behaviour, reducing social welfare. Third, norms and values may influence the set of other agents with whom an individual interacts. Values that engender tolerance of racial, ethnic, or other differences may undermine the social divisions that limit the number of potential transactors. By facilitating interactions between individuals, tolerance may increase the gains from exchange both by intensifying competition, and thereby producing greater efficiency in all trades, and by generating a larger number of transactions. Values that engender intolerance will limit both the quantity and efficiency of trade. Evolutionary game theory argues that social norms or cultural traits are propagated as individuals emulate successful behaviours that they observe in

186 Jane Friesen others.5 Community life increases the pay-off and therefore the evolutionary success of economically favourable social behaviour. By lowering the cost of information about others, community social relations facilitate the establishment of reputations, which makes behaviour that imposes economic costs on others more costly to the individual. Repeated interactions among members of communities also permit retaliation, again increasing the penalty for antisocial behaviour. Cooperative equilibria, therefore, are more likely to be established and maintained in community environments than in environments where social contact is random and anonymous. Evolutionary stable equilibria may be immune to a wide variety and number of deviations from cooperative behaviour among members of the community.6 As Dayton-Johnson (this volume) notes, as long as a sufficiently large number of individuals are behaving cooperatively, it is individually optimal to do so. However, there is a critical level of defection. Once an individual encounters a certain number of individuals who behave non-cooperatively, behaving cooperatively is no longer the individually optimal strategy, and the previous equilibrium is destabilized. As more and more individuals change their behaviour, the rate at which others encounter non-cooperative behaviour accelerates. As a result, as soon as a crucial threshold of defectors has been surpassed a social system will move quickly to a new equilibrium in which all defect. The implication is that although social equilibria are stable, and may last a long time, change can be sudden and dramatic. As McCracken (this volume) argues, higher levels of trust can increase growth rates by (1) reducing transaction costs including those associated with writing explicit contracts and litigation, (2) encouraging innovation by reducing the risk of malfeasance among partners, (3) facilitating the functioning of informal credit markets, (4) improving the credibility and therefore the effectiveness of government policy, and (5) reducing nepotism in hiring.7 The recent empirical literature investigating trust and economic growth originates with Putnam (1993). Putnam argues that differences in the economic success of different Italian regions can be explained by differences in the level of social trust. These social differences, in turn, are attributed to differences in the density of horizontal social networks.8 Is trust, perhaps, a more complex phenomenon, both in terms of its economic implications and in the social relations that generate it than is considered in these papers? The Putnam hypothesis emphasizes the role of repeated interaction and density of social networks in forming norms and values. Certainly, the relative homogeneity of segmented communities and barriers to migration across communities may support the formation of positive social norms such as honesty, cooperativeness, and charity. Trust, in this context,

Communities and Economic Prosperity

187

Figure 1 The ambiguous effects of stable communities on norms, values, and economic prosperity

Stable, homogeneous communities

Norms of reciprocity, trust

Social distance, intolerance

Low transactions costs, internalized externalities

Unrealized gains from trade

Economic prosperity

Economic losses

means reciprocal relations between a closed community of homogeneous individuals. However, while these aspects of community life may foster local trust within homogeneous communities, segregation and parochialism may foster mistrust between groups by undermining norms of tolerance. Glaeser et al. (2000) provide experimental evidence that individuals who are socially closer are more likely to trust one another and to be trustworthy, but that racial or ethnic differences reduce trust between individuals. Viewed from this perspective, the contributions to economic life of dense community relations and strong value-creating institutions are not unambiguously positive. Figure 1 illustrates this ambiguity. The left-hand branch of the schematic describes the positive economic role played by dense community relations. Associative organizations and repeated interactions may generate trust within groups. Both through repeated interactions in informal settings and within religious and educational institutions productive norms of reciprocity are promoted and instilled. These norms and values can internalize externalities and reduce a variety of transaction costs, increasing economic prosperity. While increasing trust within communities, these same associations may reduce trust across communities by reinforcing cleavages among the members of a society. Religious and educational institutions may produce values that undermine economic success and sow divisions within communities. The potentially negative economic role played by dense community relations is described by the right-hand branch of Figure 1. Increased social distance

188 Jane Friesen between communities leaves potential gains from trade unrealized by reducing the number of potential trading partners, thus weakening economic performance. Depending on the magnitudes of their effects on within-group trust and across-group cleavages, associative organizations might, on balance, make a positive or negative contribution to the productivity of community life. Like the empirical evidence cited earlier, the theoretical relationship between community organizations, trust, and economic prosperity is ambiguous. The implications of this ambiguity for the design of community institutions are discussed later on. Networks Communities may also play an important role in reducing the incidence of market failure resulting from incomplete or asymmetric information. A costless byproduct of repeated interaction among members of communities is information, as individuals learn about one another. As well as the role that this revelation may play in supporting positive social norms, the dissemination of information about individuals through social networks can increase the efficiency of a variety of economically important matching processes. Matching processes are important whenever the quality of an economic interaction is determined by the nature of the match between the people involved. If the group of people is heterogeneous, the acquisition of information about potential partners in a match can improve match quality.9 Examples of economically important matching processes include matching consumers to goods and services, matching workers to firms in the labour market, matching workers to jobs within firms, and matching firms to suppliers of intermediate inputs. Montgomery (1991) focuses on the role of information networks in labour markets when firms cannot observe the ability of workers directly prior to hiring them. This information asymmetry can be partially overcome if firms hire new employees on the basis of referrals from existing employees. Workers who refer family and friends will reveal unobservable information about the quality of potential hires to the employer in two ways. In this model, if workers are likely to refer workers similar to them, they will passively provide information about the job candidate simply by referring him or her. A less passive process can also be imagined, whereby workers endorse or vouch for family and friends to the employer. In doing so, the worker risks her reputation with the employer, and she is compensated for this risk via the social debt of the recruit.

Communities and Economic Prosperity

189

Figure 2 Ambiguous effects of 'dense' communities on social networks and labour market efficiency Stable, homogeneous communities Dense social networks

'Inbreeding'

Solution to asymmetric information problem in labour market

Nepotism in job referrals

Efficient labour markets

Discrimination, inefficient labour

The empirical literature on the use of social contacts in the job search and recruiting process summarized by Montgomery (1991) supports the importance of these referrals. Over half of all workers currently employed in the United States found their jobs through friends and relatives. Holzer (1988), Campbell and Marsden (1990), and Osberg (1993) all find that a very high proportion of employers use referrals to fill jobs. By providing at least a partial solution to the asymmetric information problem faced by firms, social networks can increase the efficiency of the labour market matching process. Montgomery shows that efficiency improves with the number of social contacts each worker has. If community organizations that bring members of a community together in repeated interactions provide important opportunities for the formation of these information networks, a richer, denser community life would be expected to increase market efficiency. This effect of social networks on economic prosperity is represented by the left-hand side of Figure 2. As well as acting as conduits through which valuable information may flow, social networks can enable collusive behaviour. Individuals with particular group affiliations may act together to exclude non-members from information about job opportunities. Like collusive oligopolists, they may attempt to extract rents from the labour market. The right-hand side of Figure 2 illustrates that community organizations may foster 'inbreeding' based on

190 Jane Friesen non-productive characteristics. Nepotistic job referrals that fail to overcome the asymmetric information problem may further strengthen the 'old boys' club.' By exacerbating information problems, this kind of nepotism would reduce labour market efficiency. The importance of gender, race, and religion in contemporary social networks and social relationships suggests that large potential gains from exchange in the labour market remain unrealized. Berger (1995) observes that, not only do significant majorities of workers use own-sex contacts in the labour market search process, but using a male contact to obtain a job on average yields a higher wage offer than using a female contact, regardless of the worker's own sex. Whether an individual characteristic such as race, religion, or gender affects a worker's labour market productivity in different matches is itself determined by the social norms and social relations within a society. In societies that are intolerant of religious differences, for example, belonging to a particular religious group may be a necessary condition for being able to work productively along with others in a particular workplace. However, if work teams benefit from having a variety of approaches to problem solving, work group homogeneity may come at a price, in lower team productivity. Community institutions, therefore, may be important in determining what sorts of matches will be productive, as well as in opening up information channels that facilitate labour market matching. Institutions that bring members of different social groups together into communities may foster more productive labour market matching. Communities that organize themselves in ways that exploit the opportunities that social networks provide to yield information about difficult-to-observe characteristics will produce better labour market matches for their members. Human Capital Spillovers As well as supporting social norms and opening up information channels to learn about one another, social contacts between members of a community afford opportunities for individuals to learn from one another. For example, social skills such as problem solving, parenting, organization, language acquisition, or information gathering may be modelled or discussed in social situations. These skills are all vital to the economic success of individuals, families, and communities as a whole. Knowledge spillovers inevitably occur during the social contact arising in community organizations such as clubs, teams, or unions. Schools are particularly important. Children learn by imitation, and social networks determine what they see to imitate. By bringing

Communities and Economic Prosperity

191

Figure 3 Ambiguous effects of 'dense' communities on human capital formation Dense social relations More human capital spillovers

Human capital spillovers more important

Higher rates of human capital formation

Increased segregation

Unequal rates of human capital formation

together children from families with different amounts and types of human capital, schools provide important opportunities for knowledge to be shared within communities. Dense social networks that increase the number of contacts between individuals will increase the number of opportunities for knowledge spillovers to occur. This relationship is represented by the left-hand side of Figure 3. Benabou (1996b) shows that when individuals benefit from the human capital of other members of their community through knowledge spillovers, all individuals have the incentive to locate in communities that are richest in human capital. If families with more human capital are more likely to perceive differences in community quality and to have the motivation to locate themselves in high-quality communities, or if capital market imperfections mean that families with more human capital are more likely to have the means to locate themselves in high-quality communities, high human capital families will congregate in exclusive communities. A minor degree of heterogeneity of individual preferences or wealth can generate large differences in human capital across communities. This process is represented by the righthand side of Figure 3. In addition to generating significantly greater social and economic inequality, greater socioeconomic segregation will result in less human capital being created in total. When a family with a lot of human capital decides to move to a better quality neighbourhood, there are positive spillovers to families in the new neighbourhood and a loss of positive spillovers to families in the old neighbourhood. The net effect on overall learning depends on whether the

192 Jane Friesen gains to the relocated family and families in the new neighbourhood are greater than or less than the losses to the families in the old neighbourhood. Unfortunately, if family decision makers base their decisions only on the effect of the move on their own family's well-being, the losses incurred by others will not figure into their decisions. The result will be a degree of segmentation that generates less human capital formation in aggregate than could be achieved with a different configuration of families and neighbourhoods. Benabou's model implies that greater geographic segregation will make more richly endowed individuals better off and the less fortunate worse off, and it will reduce the average level of human capital formation within the broad geographic area. Evidence in support of the first two propositions is provided by a growing literature that links neighbourhood characteristics to individual socioeconomic outcomes (see, e.g., Jencks and Meyer 1990; Case and Katz 1991; and Crane 1991). Borjas (1995) and Cutler and Glaeser (1997) show that ethnic segregation contributes to the persistent disparities among different groups in U.S. cities. The third proposition, that aggregate human capital formation is reduced by segregation, is supported by a small body of evidence that shows a negative correlation between economic segregation and the overall rate of growth of per capita incomes and employment in urban areas in the United States (Ledebur and Barnes 1992; Glaeser et al. 1995). The role of community organizations in the creation of human capital is again ambiguous. Dense community social relations can facilitate the sharing of knowledge. Complementarities between human capital and community capital, however, can also produce geographic segregation, which in turn undermines the universality required for the efficient production of human capital, efficient information networks, and economic exchange unfettered by social constraints. Bowles and Gintis (1998) argue that successful communities of the future will be 'structured so as to minimize the tension between the universalism required for exploiting gains from trade and the parochialism which provides surrogates for missing markets.'10 What relevance does this have for the future of Canadian communities and the nature of policies that might structure successful communities? Universalism versus Parochialism: How Can We Structure Successful Communities? The distribution of family resources is changing in several dimensions that may affect community cohesion. Children increasingly live in households

Communities and Economic Prosperity

193

with only one parent present. More parents in all family types spend more of their time working in the paid labour market, leaving less time for nonmarket activities. An increasing proportion of jobs involves hours and work schedules that vary in unpredictable ways. This trend towards contingent employment creates significant uncertainty about both the level of family income and the possibility that parents will be available to participate in regular activities related either to parenting or to participation in neighbourhood life. Finally, high rates of immigration from non-English-speaking source countries mean that many families face the additional challenges of acquiring English language skills and adapting to a foreign culture. Should Canadians be concerned? Benabou's analysis suggests that any increase in inequality across families within a geographic region will lead families who are more motivated or more able to leave these neighbourhoods and move to more affluent areas. These understandable private decisions amplify the initial increase in inequality in three ways. First, the exodus of the well-off from affected neighbourhoods may further deprive newly impoverished families of the positive spillovers generated by those who leave, magnifying the initial increase in inequality. William Julius Wilson (1987) provides a graphic description of this process in his examination of the causes of the social and economic collapse of inner city ghettos in the United States. In the Canadian context, we should consider whether new immigrants might find themselves ghettoized and cut off from exposure to native English speakers in the school system and from access to information about the ways to get ahead in their new country. Single-parent families, who have little time or money themselves, may find themselves in communities where there are too few families with time to put into school and community organizations, and no one is able to take up any of the slack. Parents with varying work schedules may find it difficult to locate other parents who can help out with child care, little league sports, or other community activities. Second, the loss of 'community quality' associated with the out-migration affects all families who remain; not only those whose private circumstances have changed. Families who have not experienced divorce, who have stable jobs, and who are native born, but who are located in neighbourhoods that are experiencing these kinds of changes, will suffer as fewer of their neighbours are able to participate in schools, provide adequate parenting, or contribute to the pool of knowledge regarding how to 'get ahead.' Third, increasing geographic segregation may perpetuate the impoverishment of all of these families, as they are cut off from job market information networks and valuable human capital spillovers. Because family structure, labour market structure, and immigration patterns can affect the learning environments of children and their access to the information and tools re-

194 Jane Friesen quired to improve their socioeconomic conditions, geographic segregation of the poor may make it more likely that the adversity faced by one generation will be transmitted to the next. Myles et al. (2000) examine changes in within- and between-neighbourhood inequality in eight Canadian cities between 1980 and 1995. They find substantial increases in between-neighbourhood inequality over this period of time, reflecting both falling incomes in low-income neighbourhoods and rising incomes in high-income neighbourhoods. These trends suggest that the ability of neighbourhoods to generate spillovers between different income groups is very limited and that it is declining in the eight cities for which measures are presented. If Benabou's model holds, this increased geographic segregation by income will amplify the effects of increased economic adversity for many families whose individual circumstances have deteriorated, widen the circle of those adversely affected, and perpetuate the losses by reducing rates of human capital formation in poor communities. Are there policies that we can implement to impede this process? If the dynamic modelling of social-norm formation and social segregation described above is correct, small initial differences in community quality can generate large degrees of segregation. An early response to changing social conditions, therefore, would be most effective. The appropriate policy response will depend on the exact nature of the adverse shock. The structure of the public school system is often singled out as an important policy lever. Children's learning has been shown to be directly affected by the socioeconomic backgrounds of classmates (Willms 2001; Hoxby 2000), and families' location choices have been shown to be sensitive to school quality (Black 1999). Small changes in the distribution of families by socioeconomic status may generate a net outflow or reduce the inflow of more richly endowed families if there is not an offsetting increase in resources to the neighbourhood school. If measures to prevent increased geographic segregation are not pursued or are not effective, communities can be reintegrated either by enticing betteroff families back into poor neighbourhoods and/or by moving poorly off families into rich neighbourhoods. To persuade richly endowed families to move to neighbourhoods with little social capital, compensation must be offered that would offset the loss of social benefits. If social capital and personal wealth are strongly complementary and jointly productive, these losses may be large, making reintegration not feasible. Although poorly endowed families may prefer to be in neighbourhoods with more social capital, they are usually prevented from moving into those neighbourhoods by high housing costs. Because neighbourhoods with more social capital, better schools, and so forth are more desirable, the greater

Communities and Economic Prosperity

195

demand for the limited amount of housing in those communities drives housing prices up. Through this market mechanism, community quality becomes capitalized into land values. To the extent that the quality of the local school is an important factor driving the demand for housing, neighbourhoods with good schools that generate significant positive externalities across students will become more expensive. High land values then function as a device that rations access to good public schools in the same way that high tuition fees do in a private school system (see, e.g., Black, 1999). The exclusion of low-income families from good schools through the real estate market can be undermined through a variety of policies. Zoning regulations that promote the construction of multiple-unit housing, housing subsidies, removal of restrictions on basement suites, and policies that assist the establishment of housing co-ops and provide greater variation in lot sizes in high-quality neighbourhoods can promote greater socioeconomic integration. Rent controls also can achieve greater accessibility to high-quality neighbourhoods in the short run, but they are likely to exacerbate the problem in the long run as the number of rental properties in these neighbourhoods declines. School catchment area boundaries could be redrawn or schools enlarged so as to generate more heterogeneous school populations. However, these strategies will succeed only to the extent that socioeconomically deprived neighbourhoods are contiguous with neighbourhoods that are significantly more advantaged and that better-off families do not leave the neighbourhood or send their children to private schools. Faced with increased segregation, governments could simply provide more resources to poor schools to compensate for the lack of social capital. However, the costs may be substantial. In Benabou's words, Top-notch teachers, computers and other educational resources may not do much in a school plagued by discipline problems, the lack of motivation from role models in the community, peer pressure not to study, neighborhood gangs, etc.'11 The benefits of moving a community over the threshold to a self-enforcing set of positive norms may also be enormous. Strategies to prevent segregation before a negative dynamic is established are likely to be more cost-effective than trying to remedy a situation after it has deteriorated. Finally, resources could be directed towards supporting and enhancing the direct role that public schools play in countering some of the disadvantages of segregation, by shaping attitudes towards diversity and values of tolerance and universality through the curriculum in all schools. High school social studies curricula are often designed with the intention of developing the willingness and ability to undertake the responsibilities of citizenship. Indeed, one of the historic objectives of public schools is to fulfil this role. If 'soft'

196 Jane Friesen subjects such as social studies, the arts, and literature play an important role in developing tolerance and a sense of broader social responsibility, a shift in emphasis away from them could weaken the social underpinnings of successful community life in segregated societies. This aspect of the curriculum may be particularly important in societies that are highly geographically segregated. The argument that public schools are focal points for the kinds of community interactions and segregation dynamics just described assumes that characteristics like the stability and cohesiveness of communities affect individual school outcomes, that individual socioeconomic characteristics generate spillovers between students, and that these neighbourhood characteristics and spillovers within schools are a driving force behind locational choice and therefore socioeconomic segregation. The next section of this chapter provides an empirical exploration of the first two of these assumptions. Neighbourhood Characteristics and School Outcomes - Some Evidence from High Schools in British Columbia The first section of this chapter discussed the potentially beneficial role of social cohesion and pointed to the possibility that school outcomes might be affected by community stability and heterogeneity. Students in more stable communities might do better in school because parents are more willing to invest in school and community-specific relationships. Evidence supporting this hypothesis is found in Glaeser et al. (2000), who report a strong negative relationship between the probability of moving and membership in organizations, and in DiPasquale and Glaeser (1998), who report a strong positive relationship between home ownership and a wide variety of citizenship variables. Parental involvement in school has been shown to have a positive effect on scholastic achievement and to reduce the within-school variance in achievement, conditional on socioeconomic characteristics (see, e.g., Ho and Willms 1969). Students in more heterogeneous schools might do better in school on average if there are positive spillovers from more advantaged to less advantaged students that outweigh the adverse effects on more advantaged students of being in a more diverse environment. A wide variety of studies have shown that streaming children based on scholastic ability improves outcomes for advantaged children and worsens outcomes for disadvantaged children (see, e.g., the many references in Willms 2001). Alesina and LaFerrara (2000) find that greater ethnic, racial, or income heterogeneity measured at the state level reduces social participation in the United States. Thus far, this chapter has outlined a number of reasons why social cohesion may have either a positive or a negative impact on socioeconomic outcomes. The crucial issue is which

Communities and Economic Prosperity

197

dominates. One can predict that academic success rates may be affected by measures of neighbourhood income inequality, sociological heterogeneity, and stability - but how? To examine this, I have constructed a model of school results that includes several measures of community stability and heterogeneity, using a data set that matches school-level performance measures in English and mathematics on Grade 12 British Columbia Provincial Examinations to a set of variables measuring a variety of socioeconomic characteristics of the census tract that each school is located in. Students in British Columbia are required to take four provincially examinable courses in order to graduate and to enter B.C.'s universities, of which one must be English, Communications, or French language. Exam grades count for 40 per cent of the final course grade, and course work counts for 60 per cent. Students who do not write the English exam must either write the Communications or French language exam, postpone graduation, or simply fail to graduate. Students who do not write the math exam choose some other examinable field, and in so doing may generate a signal to post-secondary institutions of some weakness in mathematics. The proportion of students who write the English exam, therefore, can be interpreted as a crude indicator of the proportion of students enrolled in Grade 12 who are prepared for graduation. The proportion of students who write the math exam can be interpreted as an indicator of preparedness in this demanding subject that is crucial for future academic success in a variety of post-secondary fields.12 The first rows of the upper panel in Table 1 indicate that in 1994, over three-quarters of Grade 12 students in the sample wrote the provincial English examination, compared with slightly under half who wrote the provincial mathematics examination.13 These data were matched to data from the 1991 census. The census tract variables describe the neighbourhood socioeconomic characteristics as they existed when the students whose results are being analysed were in grades 9 and 10. The census tract characteristics that were included in the model of school results are listed in the lower panel of Table 1, along with descriptive statistics. The rate of population growth in the census tract was included in the model to account for community stability. Higher rates of population change imply weaker social bonds within a neighbourhood and, therefore, fewer opportunities for a variety of information and knowledge spillovers that may complement the classroom learning process. We would expect a higher rate of population change to be inversely related to school success rates. The proportion of foreign-born persons living in the census tract was included because the strength of community links may be different for immigrants than for non-immigrants. Those born outside of Canada may not have

198 Jane Friesen Table 1 English and math results: descriptive statistics, sixty-six public high schools, Vancouver, 1993-4

Variable Academic results (%) English exam Grade 12 students who wrote it Pass rate Students graded B or better Math exam Grade 12 students who wrote it Pass rate Students graded B or better

Mean

Coefficient St. dev. Minimum Maximum of variation

77.63 89.03 34.46

11.38 5.78 10.68

54.5 76.1 11.8

107.1 100.0 58.7

0.15 0.07 0.31

47.73 83.14 41 .99

19.60

13.3 56.8 12.4

122.9* 100.0 73.6

0.41 0.11 0.32

9.46 13.25

Census Tract Characteristics -1.00 116.5 23.31 Change in population, 1986-91 (%) 20.36 Mother tongue (frequency) English 0.74 0.91 0.370 0.15 0.34 0.001 0.10 Chinese 0.08 0.44 0.002 0.12 Chinese ethnic origin (frequency) 0.10 0.140 0.00 0.59 Foreign born (frequency) 0.28 0.04 0.13 0.001 Arrived in Canada since 1986 0.05 Families with children (frequency) 0.52 0.01 0.00 Headed by single parent 0.21 0.18 0.06 Mother not in labour force 0.30 0.51 High school diploma 0.94 0.52 0.10 (adults in census tract) 0.70 Mean family income ($) 58,473 136,000 18,493 31 ,560 942 8,621 St. dev., family income 2,380.1 1,596.9

1.15 0.21 1.29 1.21 0.39 0.75 0.46 0.21 0.14 0.32 0.67

Sources: Success and Participation Rates for Grade 12 Provincial Exams by School, Report Number 5019B, British Columbia Ministry of Education; Statistics Canada, 1991 Census, Results by Census Tract. *The proportion of students who write the exam can exceed 100% for three reasons: (1) some students write the exam more than once in a year; (2) Grade 11 students may elect to write the exam; and (3) the figure is calculated as a proportion of September enrollment, whereas the exams are written later in the year when enrollment may have increased.

had the same opportunities to acquire country-specific human capital and become integrated into networks of native-born Canadians. Immigrants who are members of large ethnic communities may have strong community and cultural ties within the immigrant community that provide them with the advantages of a socially cohesive culture. People with Chinese ethnic origins

Communities and Economic Prosperity

199

comprise by far the largest foreign-born ethnic group in British Columbia. The variation in the geographic settlement patterns of Chinese-speaking and other immigrants affords an opportunity to examine the effect of high concentrations of immigrants on neighbourhood school results. Of course, immigrants with weak English language skills might be expected to perform relatively poorly compared with those for whom English is a first language, therefore, language controls are included in the model. Finally, the standard deviation of family income within the census tract is included to investigate whether more economically homogeneous or more economically diverse communities generate better exam results on average. Family income may provide a proxy for a variety of characteristics that are positively correlated with school performance and may capture differences in resources available to children within the community. Holding constant the mean level of family income in a census tract, a higher standard deviation of family income means there is greater potential for spillovers between students from more-advantaged backgrounds to those who are less advantaged and that there is greater potential for less-advantaged students to drag down the more advantaged. If the 'curvature' effects are such that the former dominates, a larger standard deviation of family income will be associated with better school results on average. A negative relationship between inequality and school results could also be interpreted in a way analogous to the relationship between inequality and population health (see Lavis and Stoddart, this volume). A common interpretation of this finding is that income inequality inhibits social cohesion, which in turn undermines population health. This same lack of cohesion may render public schools less effective through avenues like reduced parental involvement. The remaining census tract level variables listed in Table 1, measures of family structure, income, and adult's education, are included to control for links between socioeconomic characteristics and educational outcomes. Each of the dependent variables describing different aspects of the math and English exam results was regressed on the same set of variables. In all regressions the data were weighted by the number of students writing the exam in each school to ensure homoscedasticity of the error terms. The results for the English exam are presented in Table 2. Results are reported for three different dependent variables: the proportion of students in Grade 12 who wrote the exam, the proportion of students in Grade 12 who wrote the exam and passed, and the proportion of students in Grade 12 who wrote the exam and earned a B grade or better. Two or three specifications are presented in each case. The three social cohesion variables are included

Table 2 English exam coefficient estimates (two specifications for each case): Proportion writing, pass rates, and A and B rates (f-statistics below)

Population change Foreign bom St. ctev., family income

Proportion writing

Proportion writing

-0.027 (0,52)

0.002 (0.03)

-11,56 (0:33) -0.53E-03 (0.39)

-88,59 (-118) -101E-04 (-0.007)

Proportion writing

-

Mean family income

0.33E-03** -2.46E-05 . (2.59) (-0.13)

English mother tongue

~~

Chinese ethnicity

-

Single-parent family

-

-

-38.04* (-1.52) -45.8* (-1.27) -37.63** (-2.24)

UnconPass rate Pass rate (conditional (conditional ditional on % writing) on % writing) pass rate 0.063** (2.30) -34.25* (-1.75)

0.093*** (2.95) -94.78** (-2.28)

-0.42E-03 -9.64E-04 (-0.55) (-115)

UnconA or B rate A or B rate (conditional (conditional ditional on % writing) on % writing) A or B rate

0.01*** (2.91)

0.67 (163)

(172)

o.or*

0.08 (167)

-10.60*** (-2.54)

-14.04 (-0.40)

-83.80* (-139)

-104.32* (-169)

0.00 (-1.14)

-0.23* (199) 0.24** (2.25)

-0.003** (-2.11)

-0.002** (-2.06)

0.12 (1.66)

0.13** (1.73)

-

0.11E-03 (1.56)

1 .OOE-04 (0.96)

0.00 (0.91)

0.39E-03*** 2.64E-04** (1.73) (3-46)

28.81** (2.06)

24.00* (1.72)

-

35.85** (1.76)

27.03 (1.31)

63.60*** (3.18)

57.80*** (2.89)

-

69.80*** (2.40)

59.18 (120)

*~

-

-4.54 (-0.47)

-9.28 (-0.10)

:

0.23** (2-18)

10.89 (0.78)

-

0.003 (164)

2.18 (0.15)

Mother not in labour force

-

—

60.82" (1.81)

-

15.31 (0.81)

23.00 (1.23)

-

21.16 (0.77)

35.25 (1.28)

High school graduates (adults in census tract)

-

66.43"* (2.58)

-

3.13 (0.21)

11.50 (0.08)

_

35.36* (1.63)

50.75" (2.40)

45.82*** (3.93)

53.10"* (4.80)

-3.29 (0.41)

66

66

66

CONSTANT

59.25*" 10.06

57.80*** (2.91)

n ff

66

66

0.26

0.36

-

73.24*** (13.52) 66

0.23

Indicates statistical significance at the 10% level. Indicates statistical significance at the 5% level. Indicates statistical significance at the 1% level.

0.38

0.34

-

0.37

-56.09*** (-3.31) 66

0.50

-42.70** (-2.61) 66

0.46

202 Jane Friesen along with mean family income in the first. The additional covariates described earlier are included in the second. In the pass rate and 'A and B rate' equations, the model is estimated with and without including the proportion of students who wrote the exam as an explanatory variable. The population change variable exhibits a positive coefficient in all but the first model reported in Table 2. This evidence does not support the hypothesis that high growth rates are associated with less social cohesion and poorer school results. An alternative explanation is that there is positive self-selection into high-growth areas. However, immigrant density is controlled for in all specifications, so if there is a selection effect in high-growth neighbourhoods, it is not associated with the qualities of immigrants. Higher proportions of immigrants are associated with poorer English results. This effect is greater in specifications that include a measure of the proportion of people who report that their single ethnic origin is Chinese. Holding constant a number of other variables, including the proportion of residents who speak English and immigrant density, English results are better in schools located in census tracts where a greater share of residents report Chinese ethnicity, compared with census tracts where Chinese residents form a smaller proportion of the population. These results are consistent with two related hypotheses: immigrants may perform relatively poorly because they lack country-specific human capital and have relatively weak information and support networks, and Chinese immigrants, as members of the largest immigrant group in the Lower Mainland of British Columbia, have better access to information and support than other immigrant groups. A higher standard deviation of family income has no statistically significant effect on the proportion of students writing the English exam or the pass rate, but it does cause a statistically significant reduction in the proportion of students writing who earned a B grade or better. This suggests that the performance of the best students is adversely affected by being part of a more economically diverse student body, but it does not indicate that there are spillovers pulling marginal students up to a passing level. Alternatively, greater economic inequality may undermine community organization and parental involvement in schools. The English exam may capture a broad set of skills in graduating students, including the ability to comprehend and integrate conceptual ambiguities, culture-specific linguistic connotations, and social perceptiveness, as well as simple language proficiency. According to the statement of rationale for the English Language Arts 11 and 12 courses, 'Language is fundamental to thinking, learning, and communicating in all cultures. The skilled use of language

Communities and Economic Prosperity

203

is associated with many opportunities in life, including further education, work, and social interaction.'14 As the only examinable subject required for graduation (with minor exceptions for French Immersion students and a Communications option), the English exam is clearly meant to measure fundamentally important dimensions of students' education and socialization. If one aspect of social cohesion is the degree to which individuals are able to engage one another in the context of a common linguistic, cultural, and social core, the strength of students' performance on the English exam can be thought of as itself providing an important indicator of how well each cohort is prepared to contribute to a cohesive society. This perspective suggests that the English exam results (1) are affected by the degree of social cohesion in the surrounding community and (2) will affect the cohesiveness of our society in future. If we care about building a socially cohesive society, we therefore should be concerned about the socioeconomic trends that are likely to undermine the English exam results. The results for the control variables are for the most part unsurprising. Table 2 indicates that a higher proportion of adults in the census tract with at least high school education is associated with better exam performance.15 Education has the largest effect on the English results of any of the variables included. An increase of 1 per cent in the proportion of adults who have a high school education or better increases the proportion of students who get an A or B on the English exam by 1.8 per cent. Family structure also affects the English exam results. A higher proportion of mothers who are not in the labour force increases the proportion of students who write the exam, but it has no effect on the success rates. A higher proportion of single-parent families in the census tract lowers the proportion of students who write the English exam, but this variable has no statistically significant effect on the pass rate of those who write the exam or the proportion of those who write who earn a B grade or better. The results for mean family income show no effect on the proportion of students who write the exam or who pass the exam, holding constant the other population characteristics in the model. However, a higher proportion of students earns A's and B's in census tracts with higher mean family incomes, all else being equal. This effect is statistically significant when the proportion of students writing the exam is included in the model, but it is marginally insignificant when this variable is omitted. The sensitivity of the English exam results to family income, immigration, and the availability of parents is worrisome in light of the high rates of immigration into Canada's major cities, the high rates of single parenthood

Table 3 Math exam coefficient estimates (two specifications for each case): Proportion writing, pass rates, and A and B rates (^-statistics below)

Population change Foreign bom St ttev., family Income

Proportion writing

Proportion writing

-0.06 (0.74)

0,00 (0,30)

224.35*** (3.76) 0.002 (0.94)

-26,41 (-0,29)

0,00 (0,66)

Pass rate Pass rate Uncon(conditional (conditional ditional on % writing) on % writing) pass rate

-0,02 (0.66)

-0.06 (-1.42)

-0,06 (-1.43)

-0.07 (0.97)

-0,07 (-135)

-0.07 (-1.27)

92,74*** (3.18)

45.9 (0.88)

55,38 (0.98)

130.85*** (3.63)

32.32 (0.49)

31.94 (0.47)

-0.39 (0-35)

-0.00 (-0.12)

-0,00 (0.05)

0.00 (0.02) 0.07 (0.97)

0.21** (2.13)

0.00 (1.17)

0.00 (1.08)

0.00 (1.37)

0.00 (1.34)

0.00 (1.29)

-

20.44 (1.07)

15.71 (0.76)

-

-27.48 (-1.13)

-31.13 (-1.23)

-

45.52* (1.65)

37.76 (1.29)

-

-2.67 (-0.07)

9.46 (0.27)

29.99** (2.52)

32.30** (2.50)

:

38.47** (2.54)

41.32** (2.63)

-

-

-0.80 (1.42)

-.01 (-0.09)

Mean family income

0.00 (0.96)

0.00 0.22E-03** (2.20) (0.002)

English mother tongue

-

-5.2 (-0.15)

Chinese ethnicity

-

78.09* (1.65)

:

7.58 (0.36)

-4,74E-04 -0.00 (-0,36) (-0.13)

-

Proportion writing

Single-parent family

A or B rate A or B rate Uncon(conditional (conditional ditional on % writing) on % writing) A or B rate

:

-

Mother not in labour force

-

High school graduates

-

-

-

CONSTANT

21 .87" (2.05)

n FF

66

0.34

64.44* (1.78)

-

-30.81 (-1.46)

96.87*** (3.24)

-

35.30* (1.91)

36.43** (1.98)

33.14** (2.26)

38.86** (2.25)

66

66

-44.4* (-1.76) 66

0.67

-

70.32*** (14.51) 66

0.27

Indicates statistical significance at the 10% level. Indicates statistical significance at the 5% level. Indicates statistical significance at the 1% level.

0.46

-31.02 (-1.39)

0.43

-

-32.00 (-1.19)

-

37.46 (1.60)

58.38** (2.61)

23.66*** (3.94)

16.13 (0.86)

7.77 (0.41)

66

66

66

-

0.38

0.57

-18.63 (0.69)

0.54

206 Jane Friesen and two-earner households with children in school, and the increasing proportion of families with children who have low incomes. Providing resources to support individuals, families, and communities that are disadvantaged in this respect may have significant long-term consequences for their meaningful integration into a strong, coherent, and common culture. Unlike the English exam, it would be difficult to argue that the mathematics exam captures skills or qualities of students that might themselves influence their ability to contribute to the quality of their community's social fabric. Our primary interest in the math results, therefore, is to determine whether the formation of this type of human capital is affected by the stability and heterogeneity of the communities that students live in. In contrast to the English results, Table 3 indicates that none of the three variables that could lend support to the social cohesion hypothesis appear to be relevant to success rates in high school mathematics. Neither the population growth variable nor the standard deviation of income variable is significant in any of the models. The immigration variable is not statistically significant in models that include the Chinese ethnicity variable. Math results are better in schools located in census tracts where a greater share of residents report Chinese ethnicity, compared with census tracts where Chinese residents form a smaller proportion of the population. Of the remaining variables, a higher proportion of high school graduates in the census tract is associated with a higher proportion of students writing and passing the math exam. All else being equal, students are more likely to write the math exam at schools located in neighbourhoods in which there is a higher proportion among the mothers with children at home who do not participate in the labour force. Surprisingly, a higher proportion of students pass, and a higher proportion earn A's and B's in neighbourhoods with higher concentrations of single-parent families. However, these estimated effects are not large - the elasticity of the pass rate is 0.07 and of the A and B rate is 0.18. In contrast, the elasticity with respect to the education variable is 1.3 at the mean of the sample. High levels of proficiency in English appear to be much more sensitive to family and location disruptions and differences in income than proficiency in math at the Grade 12 level. Considering that the math exam measures strictly academic achievement in a narrowly defined technical subject, while we have argued that the English exam captures a broader set of comprehension, communication, and other social skills, the relative sensitivity of the English results to socioeconomic conditions is not surprising.

Communities and Economic Prosperity

207

Conclusion The empirical work presented in this chapter found that mean neigbourhood family income, ethnicity, the proportion of single-parent families, the proportion of mothers who do not participate in the labour force, and the density of recent immigrants all have significant effects on the performance of Grade 12 students on the Provincial English examination in British Columbia. It was argued that the Grade 12 English exam results can be thought of as capturing some important aspects of the degree to which the cohort of high school leavers is able to communicate effectively, think, express themselves in a nuanced way, and grasp complex social ideas. If so, the sensitivity of these results to the social and demographic factors examined here should alert us to the possibility that recent social trends, including increased dispersion and segregation of families by income, growing numbers of single-parent families (many of whom have low incomes), many families with two parents working, and high levels of immigration, will leave us with a society with a lower capacity to function cohesively in future. The theoretical and empirical literature reviewed in the first section of this chapter provides us with further reasons to be concerned about the long-term consequences of undermining the quality of the social fabric in these ways. First, there is no reason to expect a priori that the market on its own will generate a social equilibrium that produces the best possible economic outcomes. In particular, segmentation and stratification will undermine communities' abilities to produce a variety of local public goods. Greater degrees of diversity among members of a community will result in greater segmentation and stratification. Second, the evolutionary approach to modelling the dynamics of community norms suggests that our social systems may be able to withstand a lot of deviations from desirable social behaviour without upsetting good equilibria. However, once the number of 'defectors' from prosocial behaviour passes a certain threshold, the social fabric can deteriorate suddenly, dramatically, and with little warning. These invisible thresholds imply that we cannot know how much more social stress our communities can experience before these destructive dynamics become apparent. Third, once segregation has occurred it is very costly to reverse. Strategies to prevent segregation, therefore, are more likely to be effective than strategies to undo it. The tensions between parochialism and universality, and between individual choice and broader social benefits, mean that the policy issues are complex and the appropriate policy responses may be subtle. In particular, failure to note these tensions may lead us to overly idealize closely knit,

208 Jane Friesen homogeneous communities. Although these sorts of communities can create strong relationships among their own members that generate socially and economically beneficial outcomes, they also risk undermining tolerance and universalism. Rather than rebuilding the communities of the past, the challenge Canada faces is to build communities that provide the advantages of dense social networks, but integrate additional and different segments of society to promote values that will build a society that is more broadly socially cohesive. Data Appendix Sources The high school exam results analysed in this paper were taken from 'Success and Participation Rates for Grade 12 Provincial Exams by School,' Report Number 5019B, British Columbia Ministry of Education. This publication is available on the Ministry's Web site at www.bced.gov.bc.ca/exams/ standrep.htm. The Census Tract data were drawn from the 1991 Census, Results by Census Tract, prepared by Statistics Canada. These data were accessed through the University of Toronto's CHASS Data Centre Web site at www.chass .utoronto.ca:8080/economics/eco.html. Notes 1 I would like to thank Lars Osberg, Greg Dow, Stephen Easton and David Laycock for their insightful comments. Teodora Cosac and Bogdan Buduru provided excellent research assistance. Funding for this project was provided by Heritage Canada. 2 Ben-Ner and Putterman (1998: 20). 3 Barry Schwartz, cited in Ben-Ner and Putterman (1998: 7). 4 Ben-Ner and Putterman (1998: 7). 5 For particular references to game theoretic models of the type described in this section, see Bowles and Gintis (1998). See also Dayton-Johnson's chapter in this book. 6 Young (1996) provides a very accessible summary of these ideas. Sethi and Somanathan (1996) provide an interesting application of this type of model to the management of a fishing resource by a community that depends upon it. 7 The effect of trust on economic performance is discussed extensively by Knack and Keefer (1997).

Communities and Economic Prosperity 209 8 However, the small number of empirical studies that follow Putnam provide conflicting evidence on the importance of trust as a determinant of economic growth. See Helliwell (1996a, b), Knack and Keefer (1997). The measure of trust used in research based on the World Values Survey is derived from responses to the question 'Generally speaking, would you say that most people can be trusted, or that you can't be too careful in dealing with people?' 9 See Wolinsky (1995) for a discussion of matching and information problems. 10 See Samuel Bowles and Herbert Gintis (1998) and Avner Ben-Ner and Louis Putterman, eds. (1998: 228). 11 Roland Benabou (1996: 225). 12 All of the schools included in the sample are located in the Lower Mainland area, which broadly speaking includes most areas within commuting distance of Vancouver. Private schools were excluded, along with public schools that specialized in areas such as technical education, adult learning, etc. The sample used in the estimation consists of sixty-six public high schools. The schools data obtained reported examination results for the 1993-4 academic year. 13 Because of the smaller number of courses that can be substituted for English than for Math as examinable subjects, both the standard deviation in column two and the range provided in columns three and four indicate that the variation across schools in the proportion of students who wrote the Math exam is far greater than the English exam. 14 Province of British Columbia, Ministry of Education, Integrated Resource Package 1996, located at www.bced.gov.bc.ca/irp/elalll2/elatoc.htm. 15 Measures of higher educational achievement were not included because they were found to be statistically insignificant once the proportion of adults with a high school education was controlled for.

References Alesina, A., and E. LaFerrara. 2000. 'Participation in Heterogeneous Communities.' Quarterly Journal of Economics 115(3): 874-904. Benabou, Roland. 1996a. 'Heterogeneity, Stratification and Growth: Macroeconomic Implications of Community Structure and School Finance.' American Economic Review 86(3): 584-609. - 1996b. 'Equity and Efficiency in Human Capital Investment: The Local Connection.' Review of Economic Studies 63, 237-64. Ben-Ner, Avner, and Louis Putterman, eds. 1998. 'Values and Institutions in Economic Analysis.' In Avner Ben-Ner and Louis Putterman, eds., Economics, Values and Organization. Cambridge, Mass.: Cambridge University Press, 3-69.

210

Jane Friesen

Berger, Jacqueline. 1995. Were You Referred by a Man or a Woman? Gender of Contacts and Labor Market Outcomes. Working Paper 353. Princeton, NJ: Princeton University, Industrial Relations Section. Black, Sandra E. 1999. 'Do Better Schools Matter? Parental Valuation of Elementary Education.' Quarterly Journal of Economics 114(2): 577-600. Borjas, George J. 1995. 'Ethnicity, Neighborhoods and Human Capital Externalities.' American Economic Review 85(3): 365-90. Bowles, Samuel, and Herbert Gintis. 1998. 'How Communities Govern: The Structural Basis of Prosocial Norms.' In Avner Ben-Ner, and Louis Putterman, eds., Economics, Values and Organizations. Cambridge: Cambridge University Press. Campbell, Karen E., and Peter V. Marsden. 1990. 'Recruitment and Selection Processes: The Organizational Side of Job Searches.' In Ronald L. Breiger, ed., Social Mobility and Social Structure. New York: Cambridge University Press, 59-79. Case, Anne C., and Lawrence F. Katz. 1991. The Company You Keep: The Effect of Family and Neighborhood on Disadvantaged Youths. Working Paper 3705. Cambridge, Mass.: National Bureau of Economic Research. Crane, Jonathon. 1991. 'The Epidemic Theory of Ghettos and Neighborhood Effects on Dropping Out and Teenage Childbearing.' American Journal of Sociology 96(5): 1226-59. Cutler, David M., and Edward L. Glaeser. 1997. 'Are Ghettos Good or Bad?' Quarterly Journal of Economics (Aug.): 827-72. DiPasquale, D., and E. Glaeser. 1998. Incentives and Social Capital: Are Homeowners Better Citizens? Working Paper 6363. Cambridge, Mass.: National Bureau of Economic Research. Fernandez, Raquel, and Richard Rogerson. 1996. 'Income Distribution, Communities and the Quality of Public Education.' Quarterly Journal of Economics 111(1): 135-64. Glaeser, Edward L. 2001. 'The Formation of Social Capital.' Canadian Journal of Policy Research 2(1): 34-40. Glaeser, E., D. Laibson, J. Scheinkman, and C. Soutter. 2000. 'Measuring Trust.' Quarterly Journal of Economics 115(3): 811-46. Glaeser, E., J. Scheinkman, and B. Sacerdote. 2000. 'What Is Social Capital?' Mimeo. Glaeser, E., J. Scheinkman, and A. Shleifer. 1995. Economic Growth in a CrossSection of Cities. Working Paper 5013. Cambridge, Mass.: National Bureau of Economic Research. Helliwell, John F. 1996a. Economic Growth and Social Capital in Asia. Working Paper 5470. Cambridge, Mass.: National Bureau of Economic Research.

Communities and Economic Prosperity

211

- 1996b. Do Borders Matter for Social Capital? Economic Growth and Civic Culture in U.S. States and Canadian Provinces. Working Paper 5863. Cambridge, Mass.: National Bureau of Economic Research. Helliwell, J., and R. Putnam. 1995. 'Economic Growth and Social Capital in Italy.' Eastern Economic Journal. 21(3): 295-307. Ho, E., and J. Douglas Willms. 1969. 'The Effects of Parental Involvement on Eighth Grade Achievement.' Sociology of Education 69: 126-41. Holzer, Harry. 1988. 'Search Method Use by Unemployed Youth.' Journal of Labor Economics 6: 1-20. Hoxby, Caroline. 2000. Peer Effects in the Classroom: Learning from Gender and Race Variation. Working Paper W7867. Cambridge, Mass.: National Bureau of Economic Research. Knack, Stephen, and Philip Keefer. 1997. 'Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.' Quarterly Journal of Economics (Nov.): 1251-88. Jencks, D., and S. Mayer. 1990. 'The Social Consequences of Growing Up in a Poor Neighborhood.' In L. Lynn and M. Mcgreary, eds., Inner City Poverty in the United States. Washington DC: National Academy Press. Ledebur, L., and W. Barnes. 1992. 'City Distress, Metropolitan Disparities and Economic Growth.' National League of Cities Research Report. Manski, C. 1993. 'Identification of Social Effects: The Reflection Problem.' Review of Economic Studies 60: 531-42. Myles, J., G. Picot, and W. Pyper. 2000. Neighborhood Inequality in Canadian Cities. Working Paper 160. Ottawa: Statistics Canada, Analytical Studies Branch. Montgomery, James D. 1991. 'Social Networks and Labor-Market Outcomes: Toward an Economic Analysis.' American Economic Review 81(5): 1408-17. Osberg, Lars. 1993. 'Fishing in Different Pools: Job Search Strategies and JobFinding Success in Canada in the Early 1980s.' Journal of Labor Economics 11(2): 348-86. Putnam, Robert. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Rusk, D. 1993. Cities Without Suburbs. Baltimore: Johns Hopkins University Press. Schwartz, Barry. 1993. 'On the Creation of Value.' In Michael Hechter, Lynn Nadel, and Richard E. Michod, eds., The Origin of Value. New York: Aldine de Gruyter, 153-86. Sethi, Rajiv, and E. Somanathan. 1996. The Evolution of Social Norms in Common Property Resource Use.' American Economic Review 86(4): 766-88. Willms, J. Douglas. 2001. 'Three Hypotheses about Community Effects about Social Outcomes.' Canadian Journal of Policy Research 2(1): 53-62.

212

Jane Friesen

Wilson, William Julius. 1987. The Truly Disadvantaged. Chicago: University of Chicago Press. Wolinsky, Asher. 1995. 'Matching, Search, and Bargaining.' In David K. Levine and Steven A. Lippman, eds., The Economics of Information, vol. 1; International Library of Critical Writings in Economics, vol. 53. Aldershot, UK: Elgar. Young, H. Peyton. 1996. The Economics of Convention.' Journal of Economic Perspectives 10(2): 105-22.

Social Cohesion and Macroeconomic Performance M.C. McCRACKEN1

There is increasing recognition that there is something called 'social capital,' 'social cohesion,' or 'social capability,' which affects the performance of nations. Some stress the importance of a 'social compact' or 'social contracts' among major groups - business, labour, government, and citizens (Reich 1998). Others see an explanation in the existence of strong interpersonal linkages through families, associations, and trust-building institutions (Putnam 1993). John Harriss and Paolo de Renzio (1997: 932) have constructed the following typology for social capital: • Family and kinship connections • Wider social networks, or 'associational life' - networks of civic engagement • Cross-sectional linkages or contacts spanning differences in sector or power • Political capital - the informal institutional arrangements for mediating conflict • Institutional and policy framework - the set of formal rules and norms (constitutions, laws, regulations, and policies) that regulate public life in society • Social norms and values Michael Woolcock (1998: 162) has broadened the notion of social capital by making a useful distinction between the macro- and microlevels of 'embeddedness' and 'autonomy.' Embeddedness implies that 'all forms of exchange are inherently embedded in social relationships' (1998: 163). At the same time, autonomous ties to other social entities are needed to coordinate

214 M.C. McCracken outcomes. Embeddedness at the micro level refers to intra-community ties, whereas at the macro level it refers to state-society relations; autonomy at the micro level refers to extra-community networks, while at the macro level it refers to institutional capacity and credibility' (1998: 164). This chapter takes a broad view of the idea of 'social cohesion.' Its objective is to offer some insights into where one might look for a connection between the social system and macroeconomic performance. At this stage of the literature, identification of such connections would be a helpful first step, as the empirical studies that might allow quantification of the linkages are at an early stage of development. If there are linkages from social cohesion to macroeconomic performance, the question immediately arises as to which dimensions of economic performance are most affected, and in what direction? In principle, social cohesion might affect: The productive potential of the economy - that is, the potential output, at any point in time, and the production function relationship between outputs and inputs The extent to which the economy falls short of potential - that is, the output gap, at any point in time, between potential and actual output The amount of economic surplus - that is, discretionary income or surplus, reflecting the difference between total production and that required for some basic level of private and public goods and services The distribution of income (before and after transfers and taxes) of discretionary time and of health across the population The rate at which potential output grows over time; other dimensions of dynamism are the pace of innovation, the speed of convergence of regions, the spread of new products and practices, the dissemination of information, and the confidence of the citizens in the future The production function framework of Irma Adelman includes explicitly a vector of variables 'representing the entire social, cultural, and institutional complex of society' (Adelman 1961: 13). Indeed, such socioeconomic factors are seen to serve 'the role of prime movers in initiating economic development' (1961: 147). The implication is that evidence of an adverse impact of declining social cohesion would show up as a disappointing performance of total factor productivity (TFP) in a production function that includes only labour and capital inputs. Similarly, social cohesion can be considered to be a possible explanatory variable in each of the other frameworks or areas. For example, the Lavis and

Social Cohesion and Macroeconomic Performance

215

Stoddard chapter in this book examines the linkages between social cohesion and health outcomes, which may help to explain increases in health costs through declining social cohesion. Phipps's chapter discusses the impacts of social cohesion on child outcomes, and Friesen's chapter presents an analysis of educational outcomes. In each case, the implication of declining levels of social cohesion is that society will get less desirable outcomes, even if there is the same level of expenditure on labour and capital inputs - that is, total factor productivity will fall and potential output will decline. In the following section the emergence of public interest in social cohesion is examined. The next section focuses on the links between social cohesion and macroeconomic performance, followed by a section that considers a number of possible measures of social capital or cohesion for Canada and their trends over the last twenty years. Possible policy directions are then considered, followed by a brief conclusion. Social Contract or Compact How are the relationships among people 'worked out' in a society? Relationships between groupings of people may be along lines of 'opinion' and subjective identity (ethnicity, religion, and colour) or on the basis of material opportunities (labour versus capital, nobility versus 'commoners'). Indeed, the modern pluralist society is likely to be some mixture of these factors - a segmented society with a class society (Netherlands Scientific Council for Government Policy 1997: 17). When thinking about large networks within a society and the nature of the relationships between them, the notion of a 'social contract' or 'social compact' can be a useful organizing framework, with a mix of explicit and implicit contracts between groups. Indeed, some would say that the social compact could define the society or culture. In the Canadian context, the period from 1950 to 1975 had a number of reinforcing features. It was expected that successful companies would share benefits with workers by offering security of employment and a wage structure that increased with productivity gains. Firms recognized the need for stakeholder involvement and benefit and governments provided a framework of social insurance through unemployment insurance, with broad coverage, social security and social assistance. The output necessary to support these programs was there because the state accepted a responsibility to maintain low unemployment and used low interest rates to that end, while the productivity of the economy was maintained by the expenditures necessary to ensure a healthy and well-educated population. Conflict was minimized by explicit

216 M.C. McCracken attention to regional redistribution and cooperation with the provinces. Voluntary associations were nurtured, and there was an attempt to build consensus organizations like the Economic Council of Canada. The 'new state,' which is what the period beginning in about 1980 could be termed, has seen a corporate emphasis on downsizing, declining real wages, loss of benefits, anti-union activities, contingent work, polarization of wages, and an emphasis on increasing shareholder value only. Government cutbacks have affected all parts of the social insurance package, particularly Employment Insurance (El), with a lower percentage of the unemployed qualifying and with those who qualify receiving lower benefits. In old age security there has been an increase in premiums, and a clawback of the Old Age Security (OAS). The Canadian Health and Social Transfer (CHST) era has seen both provincial cutbacks and workfare on the social assistance side, together with reductions in the health care system. Rising tuition fees have undermined accessibility of higher education. Federal-provincial relations have been marked by the abandonment of regional development as an objective and downloading on provinces. The government sector has been offloading on voluntary associations and eliminating or ignoring consensus inputs. The fundamental framework has been a focus on low inflation to protect asset holders, achieved by maintaining a high unemployment rate through high real interest rates. The general label for this transformation has been the 'conservative revolution' (Allen and Rosenbluth 1992: 6ff.), although with the more recent application of more of the same policies, it now can be described as the 'neoliberal paradigm.' Why Is There an Emerging Interest in Social Capital/Cohesion? Efforts to reduce social spending are leading to dissatisfaction with governments, loss of confidence in institutions, and the emergence of a 'meaner' society. At the same time, economic performance is lower than expected, as productivity gains are weak, investment intentions poor, and demand soft. Some explanations focus on the impacts of new technologies on employment; others see globalization as the source of difficulties. Are we in a downward spiral, with economic shocks destroying social cohesion, which, in turn, is reducing productivity and increasing social tensions with subsequent costs to the economy and governments? Is it possible to reverse the spiral, and to turn it into a self-reinforcing positive virtuous circle, with rising incomes, improved social cohesion, and an improved longerterm dynamic in the society?

Social Cohesion and Macroeconomic Performance 217 The link between social capital or social cohesion and economic growth has been discussed by many in recent years. The breakdown in the social compact in the United States has been analysed by Robert Reich (1998), with a strong urging for corrective actions. Robert Putnam (1993; 1995; 1996) has stirred up interest in the concept of social capital through several of his articles. Francis Fukuyama (1995) has stressed the role of trust in explaining the varying fortunes of countries. In official circles, the Dutch government's scientific advisory group has examined the concerns about a growing social dichotomy between those who are working or not, and those with skills or not (Netherlands Scientific Council for Government Policy 1997). In New Zealand, efforts are under way to examine how social cohesion can be improved through the Foresight Project. In the project's terms, 'If social cohesion describes a society where different groups and institutions bind together effectively despite differences, then social capital must be part of the glue which enables this outcome' (Blakeley 1997). The World Bank has created a web site, focused on social capital,2 with growing traffic from both developed and developing countries. The Organization for Economic Cooperation and Development (OECD) and some member countries are discussing the concept of social capital (OECD 1998). In Canada, the Policy Research Secretariat (PRS) has identified social capital as one of the key areas for cross-departmental analysis (PRS 1996). Social cohesion or a social compact is thought by all these authors to be an important determinant for the future evolution of the economy and society. There is also a strong suspicion that we have somehow damaged the cohesion that existed in the past, with a set of actions (and inactions) since the mid1970s. But we are still left with little clarity about the mechanisms at work and how to fix the damage. This chapter focuses on the links between social cohesion and macroeconomic performance, the evidence of deterioration in Canada, and the empirical work that can be drawn on to suggest the possible pay-offs from directing more attention to Canada's social fabric. Social Cohesion Links to Macroeconomic Performance Better economic performance improves social cohesion, particularly if gains are shared among the economic partners and social programs help fill in the gaps by providing trampolines for those who need assistance to regain their place in the social structure. As well, we have seen the corrosive effects on social cohesion because of cutbacks in social programs and reduced economic gains provided to employees by employers.

218 M.C. McCracken One view of social cohesion is as a fringe benefit or 'luxury' of a wellperforming economy. This perspective argues that social cohesion is subject to cutbacks in tough economic times and suggests that there are no long-term consequences of sacrificing social cohesion (Stanley 1997: 5). However, there is growing evidence that there is a strong causal linkage from social cohesion to macroeconomic performance. The new view is that 'growth is affected by public goods, finance, demographic parameters, income distribution, and social norms, all of which are grist for the social cohesion mill' (Stanley 1997: 5). Knack and Keefer Stephen Knack and Philip Keefer (1997: 1252 ff.) identify a number of possible channels through which an improvement in social cohesion (social capital) can affect macroeconomic performance, directly and indirectly. These include: Reduced transaction costs. Trust between organizations and individuals allows transactions to take place with high confidence that, for example, payments will be made and goods shipped. Without such trust, payments are made in advance, shipments only occur after cheques have cleared, contract and litigation costs are higher, and bribes are necessary. As well, some desirable transactions do not happen at all because of the added cost and uncertainty introduced by a lack of trust. Higher investment ratio. An increase in the investment ratio occurs because greater trust and more credible policies can influence the longterm growth prospects of the nation by raising the rate of productivity growth. This works both through a higher capital-labour ratio and through a direct effect on total factor productivity (Scott 1989: 285 ff.). Encouragement of innovation. More entrepreneurial time is available for innovation, and less time is needed for close supervision of employees, partners, and suppliers. Better performance of government institutions. This includes education, the provision of public goods, better public policy, higher-quality schools (more investment in human capital); in addition, the credibility of government allows more effective policy application. Lower social costs. Lower health costs, lower police costs, and other social costs result in lower taxes or user fees, and fewer private resources to manage risks. Reductions in absenteeism and sickness improve productivity.

Social Cohesion and Macroeconomic Performance

219

• More information and accountability. Informed citizenry and accountable politicians and bureaucrats will result. All of these channels (and more) represent an expanded view of the determinants of growth beyond the simple model of labour, capital, and exogenous technical progress. These influences can reduce the costs and risks of investing, a pro-growth outcome. The measure of trust used by Knack and Keefer (1997: 1256) is the percentage of positive responses, relative to the total of positive and negative responses, to the following question in the World Values Survey (WVS) for 1981: 'Generally speaking, would you say that most people can be trusted, or that you can't be too careful in dealing with people?' In a similar vein, the strength of norms of civic cooperation are calculated from responses to a suite of questions about the justification of certain behaviours: Claiming government benefits to which you are not entitled Avoiding a fare on public transport Cheating on taxes if you have a chance Keeping money that you have found Failing to report damage you have done accidentally to a parked vehicle In the Knack and Keefer equations (1997: 1260 ff.) explaining growth across countries for the period 1980 to 1992, variables are added for trust and civic cooperation to an equation with the level of output in 1980, school enrolment rates in 1960, and investment goods prices relative to U.S. prices. Note that the explanatory measures are usually for points in time before the period for growth under examination, thereby avoiding problems of simultaneity and testing the causality flow. An increase of ten percentage points in trust increases the growth rate by 0.8 per cent; a four-point rise in the 50-point civic cooperation scale would raise growth by one percentage point. The trust and civic cooperation measures for several countries in 1981 are tabulated in Table 1 from Knack and Keefer (1997: 1285). An increase of 0.5 per cent per year in economic growth will mean 5.1 per cent more gross domestic product (GDP) in ten years, 10.5 per cent more in twenty years, and 65 per cent more in 100 years. But can we expect as much as 0.5 per cent per year from improved social cohesion? Or has the decline in the past been that much or more?

220 M.C. McCracken Table 1 Trust and civic cooperation measures by country, 1981 (%) Country

Trust

Civic cooperation

Norway Sweden Canada United States United Kingdom Germany France

61.2 57.1 49.6 45.4 44.4 29.8 24.8

40.75 41.57 39.74 40.55 40.07 39.83 36.26

If one accepts the Knack and Keefer results, a movement in Canada from the 1981 levels of trust to those of Norway or Sweden would provide more than enough of a change to yield an increase of 0.5 per cent in economic growth or $80 billion in twenty years. How can we make such a move? From a policy perspective, the challenge is to link actions of governments to changes in trust and civic cooperation.3 Social Capability and Economic Growth Another provocative study, by Temple and Johnson (1998), has linked 'social capability' to economic growth, with social capability defined as 'the attributes and qualities of people and organisations that influence the responses of people to economic opportunity, yet originate in social and political institutions' (1998: 966). These authors go back to the work of Adelman and Morris (1967), who studied the interaction of economic and non-economic forces in the course of development; they note that having the information contained in the social development index developed from data over the period 1957 to 1962, would have been a significant explanatory variable for the cross-country differences in economic growth from 1960 through 1985. One variable, communications, which measures the extent of mass communications (daily newspaper circulation per capita and radios per capita), is particularly important in explaining total factor productivity (TFP) growth. A change in the communications variable of one standard deviation change is associated with a one percentage point increase in the TFP rate (Temple and Johnson 1998: 980). Communications is highly correlated with trust, as measured by Knack and Keefer. In particular, newspaper circulation per capita is highly correlated with the trust variable.

Social Cohesion and Macroeconomic Performance

221

Temple and Johnson (1998: 987) note that 'society is important beyond its role in determining the fertility rate and the extent of investments in physical and human capital. It may be that society matters because it influences the quality of investment, the level of overall technical efficiency, or the ability of countries to assimilate technology from abroad ... an assessment of mass communications, given the absence of other good measures, is probably the best way to capture variation in social capital across developing countries.' The Role of Income Distribution Lars Osberg (1995) notes, in reviewing the recent literature on endogenous growth, that those countries with greater equality in incomes grow faster. How is this to be explained? One channel is through differing levels of human capital, with more equality of incomes encouraging higher levels of human capital formation. Robison and Siles (1997: 7-11) argue that social capital allows the realization of gains from transactions between individuals through the internalization of externalities. These gains show up as higher incomes and less income disparity. In this framework, it is the increased social capital that leads to or causes less income disparity, as well as higher income. The measures used to proxy social capital (Robison and Siles 1997: 20-3) include: Family integrity (percentage of single-family households with children under 18, birth rates of single teens, infant mortality rates) Educational achievements (high school graduation rates and teens not in school) Crime (litigation rates and violent death rates for teens) Labour force participation rates and childhood poverty rates Transfer payments for health purposes Stiglitz and Furman (1999) examine the linkages between income inequality and economic growth, through four possible channels: savings rate differentials, agency costs, fiscal policy, and political instability. Their net conclusion is that 'inequality is neither necessary for growth nor is it an inevitable consequence of growth' (Weiner and Montom 1998: 6). Richard Wilkinson suggests that social cohesion may also increase with a more equal distribution of income, as well as contributing to it. He concludes, 'So however cohesive societies are formed, a narrow distribution of income is a necessary condition for their survival and is likely to serve as a marker for important characteristics of the social fabric' (Wilkinson 1996: 135).

222 M.C. McCracken Social Institutions Ritzen et al. (2000: 6) define social cohesion as 'a state of affairs in which a group of people [delineated by a geographic region, like a country] demonstrate an aptitude for collaboration that produces a climate for change.' They go on to say 'that societies with a larger share for the middle class and more linguistic homogeneity have more social cohesion and thus better institutions, and that these better institutions lead in turn to higher growth' (2000: 21). Social institutions are both a reflection of the level of social cohesion in the society that produced them and an important influence on future levels of social cohesion. In the 1960s, a number of institutions were developed that focused Canadians' attention on longer-term issues. The Economic Council of Canada, for example, emphasized economic goals and policy recommendations based on a consensus of major stakeholders. The Science Council of Canada also undertook to develop a strategy for science policy in a number of areas. The Economic Council provided a place for Canadian leaders from the private sector (e.g., business, labour, and agriculture) to become informed about the key challenges facing Canada, to identify goals for the economy, and to review studies and make recommendations to various partners in society. The process of discussion and consensus building was aimed at developing a common perception of the constraints facing the Canadian economy and at building links between the leaders of major sectors. It worked well until 1975. The frictions around the introduction of wage and price controls in 1975 - which were introduced without consultation - led to some labour members resigning. The Economic Council then became a research organization, although still maintaining a consensus process among the remaining members. The Economic Council provided a venue for medium-term research that was not under the direct control of the government of the day, but which tried to influence the perceived framework for policy alternatives. As a result, it came under increasing pressure in the 1980s and was finally eliminated by the federal government in 1991. In the case of the Economic Council, this silenced potential criticism of the federal government's strategy of mass unemployment as an inflation control device. The willingness of the federal government to tolerate high unemployment has been very costly in lost production, and it has created a negative feedback loop by lowering social cohesion and lowering productivity growth, resulting in an even worse inflationunemployment menu in subsequent years.

Social Cohesion and Macroeconomic Performance

223

Social cohesion (or the lack of it) depends on the history of social and political relationships. In Canada, that history includes a period in which consensus-building institutions were seen as important. Is it possible to recreate some of these institutions? How else might trust in governments be restored? Measures of Canada's Social Cohesion Although we talk about the decline in social cohesion or social capital, what evidence is there that there has been any such decline? When did it start? Has it been continuous? Where do we look for such evidence? Time-use diaries indicate that most people actually spend a very small part of their day in civic or voluntary activities (Frederick 1992). Paid work (for those employed full-time) is the dominant networking or socializing activity outside the family. Exclusion from paid employment, therefore, is likely to be an important determinant of social capital. During the period from 1975 on, the unemployment rate in Canada has been drifting upwards, with several cycles. Productivity growth has slowed compared with the early post-Second World War period. Real disposable income per household has stagnated since the early 1980s. The distribution of income has worsened. Table 2 summarizes a number of social cohesion indicators and their trends from 1975 to 1995. The message is clear. By any measure, there has been a deterioration from the mid-1970s on, with little evidence of a reversal in recent years. What Could or Should be Done? As the previous section has noted, poor economic performance has coincided with a period in which a number of indicators of social capital, social cohesion, or social capability also seem to have deteriorated. The thrust of much recent economic policy has been to exclude individuals from socially valuable networks. At the same time, consensus-building institutions have been eliminated. This correlation is not accidental, and declining social cohesion could be contributing to the worsened economic performance. Can governments do anything to improve social capital? Is there any real will to do so? Some recent observations on the issue include those by Fukuyama, who states, 'Social capital ... can be dissipated by the actions of governments much more readily than those governments can build it up again' (1995: 362). And in the words of Robert Reich, 'The social compact is a promise we made to one another, and we are not keeping it when we can most afford to' (1998: 17).

224 M.C. McCracken Table 2 Indicators of social cohesion Measure

Cyclical?

Change 1975-95

Employment ratio3

Yes

Full-time youth employment ratio Bureaucratic burden: managerial & admin, share of total employment" % of employed with desired employment status Unemployed over 6 months Welfare share of population Marriages per 1 ,000 population0 Daily newspaper circulation"1 Voter turnout, federal elections Lawyers per 1 ,000 population6 % of people living in a family Unattached living with others

Yes, plus trend No

43%-38% for full-time 43%-28% 7%-14%

No

10^point drop for females; 3 point drop formales 16%-32%of total unemployment 5%-9%

No

8.2-5.2

No

0.28-0.22

No

77% (1972)67% (1997) 1.7(1985)2.2(1996) 89% (1976)83% (1997) 40% (1976)28% (1997)

No Yes

No No No

Notes

Inverted measure

Inverted measure Inverted measure No. per person (age 15+) Inverted measure

Notes: a Fortin (1996). b Gordon (1996: 42 ff). c Pullinger (1998: 8). d Temple and Johnson (1998: 977). e Knack and Keefer (1997: 1262).

Doing something about social cohesion and social capital is increasingly seen as a proper responsibility of government. However, some commentators go much further and argue that social capital is an 'antidote' to poor governments, or even a substitute. As Alejandro Fortes and Patricia Landolt (1996: 21) point out, the term 'social capital' has sometimes been stretched and

Social Cohesion and Macroeconomic Performance

225

distorted in meaning. For example, the call for more social capital is a misdiagnosis of the problems of inner cities; more economic resources are needed, not more social capital. Nevertheless, it is still useful to ask what Canada can do to rebuild social cohesion. The Position of the OECD The OECD has produced a number of studies on social programs, with a link to the fiscal difficulties of the member countries. They try to sell the view that fiscal restraint is a necessary condition for policies to promote social protection and well-being (1997: 10). They seem surprised, however, that the 'economic strategy of prudent macroeconomic policy, including fiscal consolidation, structural reform and market liberalisation' has resulted in 'witnessing in some OECD countries signs of social unrest and protest movements as well as popularity ratings for many governments and a "feel bad" factor which may reflect a lack of confidence in these policies' (1997: 9). Although there are many good intentions spelled out, one is left with the feeling that the OECD governments are not prepared to change their fiscal strategy at all, leaving little hope for much coordinated progress among the OECD countries. This suggests that we had better focus our attention on what we can do in Canada, rather than waiting for the OECD to provide guidance. This is especially the case if the OECD position represents a lagged consensus of the member countries, with particularly strong influences from the finance ministry officials in each country. Improved Institutions At one time (1964-75), Canada was well served by an institution that brought together the various private sector interests (e.g., business, labour, associations, and education), defined the economic goals of the society, and the medium-term policies with which to approach them. This served as a framework for the social compact and provided some coherence to governments in their programs. The withdrawal of labour in 1975 changed the Economic Council into a research institution and reduced its role as a bridging institution. The subsequent elimination of the Economic Council (and the Science Council) in 1991 has left no bridging vehicle for discussion of economic goals, independent from the political documents of the government of the day. Nor are there as many objective economic studies to be referred to in the policy discussions among non-government organizations (NGOs) and gov-

226 M.C. McCracken ernments. The decision to close the Economic Council had more to do with silencing public debate than showing some 'fiscal prudence,' particularly in light of the subsequent draconian fiscal and monetary policies that produced the longest recession since the Second World War (to date). ECONOMIC AND SOCIAL COUNCIL

A new institution should be developed along the lines of the Netherlands Economic and Social Council. It should be a consensus-forming organization with representatives from labour, big business, small business, provincial governments, the federal government, and NGOs (public interest). Supporting staff should include sociologists, economists, and such other disciplines as necessary to ensure a broadly based, transdisciplinary approach to the study of issues and policies. One interesting feature of the Dutch model is the requirement for the deputy minister of finance and the governor of the Netherlands Central Bank to attend meetings, along with representatives of various social policy departments. This ensures a full understanding among all parties of the macroeconomic context within which social and structural policies are pursued. SCIENCE AND ENVIRONMENTAL COUNCIL

Society is also concerned about science, technology, the environment, and the regulation of technological change. This institution would be a useful 'rebirth' of the Science Council of Canada. Its mandate should be based on an open approach to the discussion of science matters, including both an educational element and the promotion of expert discussion on key issues such as biotechnology, climate change, health, and genetics. Improved Fiscal and Monetary Policy There is little doubt that a move to stimulative fiscal policy and monetary policy could reduce the output gap, lower unemployment, improve the distribution of income, and improve the confidence of Canadians in the economy and governments. As part of such a package, reform of wage bargaining and incomes policies are an integral part. There is a need for institutions to develop a consensus about an appropriate nominal wage rate, for example. Layard et al. (1994: 98) suggest three important elements: (1) an informed national debate about what rate makes sense; (2) reports by respected and trusted bodies such as councils of economic advisers and research institutes; and (3) national-level talks between employers and unions.

Social Cohesion and Macroeconomic Performance

227

The Scandinavians have achieved a better inflation-unemployment relationship because of their corporatist central bargaining on wage levels. The alternative strategy of mass unemployment as an inflation control device is very costly in production terms, and it also creates a negative feedback loop in lower future social cohesion, lower productivity growth, and an even worse future inflation-unemployment menu. In Canada, with much of organized labour operating at the provincial level, there is the additional challenge of achieving consensus within each province. But the alternative, as witnessed by several decades of substandard performance, should be a strong motivator. WHY BOTHER?

What Is the Impact of Increasing Economic Growth by 0.5% per Year? Although this sounds quite modest, over a ten-year period this would mean a 5 per cent larger economy; over twenty years more than a 10 per cent gain. Conveniently, this is enough to close the output gap that has opened up from the early 1970s, which has also produced much higher unemployment rates than necessary and wasted public and private resources in the process. If routed to real disposable income, faster growth would turn around the declining real disposable income per household that has characterized the period from 1990 on (in 1997, average incomes were below the 1981 level by about 6 per cent). If focused on families with children, the poverty gap of about 0.75 per cent of GDP, could be eliminated in less than two years, thus ending child poverty. What Are the Advantages of a Society in Which All Persons Feel Included? For those 'outside,' the gains of living in an inclusive society are obvious. But for those currently 'inside' there are also gains, including more security of person and property, the benefits of a larger functioning society, and less of a need for redistribution of income, given full access to all. What Is the Importance of Demonstrating to the World that There Is a More Sensible Model of Social and Economic Improvement? By developing a society in which improvements in social cohesion and economic prosperity go hand in hand we can provide a working example for the rest of the world. This will help to create some sense of hope for the poorest 5 billion whose needs are very immediate, but who also need some hope for a more humane future.

228

M.C. McCracken

Notes 1 Earlier versions of this paper were presented to the Conference on the State of Living Standards and the Quality of Life (CSLS) (30-1 October 1998) and the Policy Research Committee's National Conference: Policy Research: Creating Linkages (1-2 October 1998). Comments from participants were helpful. Subsequently, detailed criticisms from Brian MacLean and Lars Osberg were quite important in clarifying and focusing the paper. Residual problems are those of the author. 2 See web site www.worldbank.org/poverty/scapital. 3 As well, the World Values Survey is conducted infrequently, with some doubt about its comparability over time (Helliwell 1998: 16). This suggests that trust may not be the 'performance indicator' for evaluating government actions.

References Adelman, Irma. 1961. Theories of Economic Growth and Development. Stanford: Stanford University Press. Adelman, Irma, and Cynthia Taft Morris. 1967. Society, Politics, and Economic Development. Baltimore: John Hopkins University Press. Allen, Robert C., and Gideon Rosenbluth, eds. 1992. False Promises: The Failure of Conservative Economics. Vancouver: New Star Books. Blakeley, Roger. 1997. Social Capital and Public Policy Development. New Zealand Minister, (as summarized in the Social Capital Database at the World Bank Web site). Fortin, Pierre. 1996. 'The Great Canadian Slump.' Canadian Journal of Economics 29(4): 761-87. Frederick, Judith A. 1995. As Time Goes By - Time Use of Canadians. Ottawa: Statistics Canada, General Social Survey 1992, 89-544E. Fukuyama, Francis. 1995. Trust: The Social Virtues and the Creation of Prosperity. Hammondsworth: Penguin Books. Gordon, David M. 1996. Fat and Mean. New York: Free Press.a Harriss, John, and Paolo de Renzio. 1997. '"Missing Link" or Analytically Missing? The Concept of Social Capital, An Introductory Bibliographic Essay.' Journal of International Development 9(7): 919-37. Helliwell, John F. 1998. 'Do Borders Matter for Social Capital? Economic Growth and Civic Culture in U.S. States and Canadian Provinces.' Paper presented at conference: Policy Research: Creating Linkages, Ottawa, 1 Oct. Knack, Stephen, and Philip Keefer. 1997. 'Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.' Quarterly Journal of Economics (Nov.): 1251-88.

Social Cohesion and Macroeconomic Performance 229 Layard, Richard, Stephen Nickell, and Richard Jackman. 1994. The Unemployment Crisis. Oxford: Oxford University Press. Netherlands Scientific Council for Government Policy (WRR). 1997. Social Dichotomy in Perspective, 50th Report. The Hague: WWR, Sdu Uitgevers. OECD. 1997. Family, Market, and Community. Paris: OECD. - 1998. The Caring World: An Analysis. Paris: OECD. Osberg, Lars. 1995. 'The Equity/Efficiency Trade-off in Retrospect.' Canadian Business Economics (Spring): 5-19. Policy Research Secretariat (PRS). 1996. 'Canadian Identity, Culture and Values: Building a Cohesive Society.' Portes, Alejandor, and Patricia Landolt. 1996. 'The Downside of Social Capital.' American Prospect 26 (May-June): 18-21, 94. Web site www.epn.org/prospect/ 26/26-cnt2. Pullinger, John. 1998. 'The Income and Wealth of Society: A System of Social Contributions.' Paper presented at the conference of the International Association for Research in Income and Wealth, Cambridge, UK, Aug. Web site www.ons.gov.uk/iariw Putnam, Robert D. 1993. 'The Prosperous Community: Social Capital and Public Life.' American Prospect 13 (Spring). Web site www.epn.org/prospect/13/ 13putn.html. 1-8. - 1995. 'Bowling Alone: America's Declining Social Capital.' Journal of Democracy 6(1): 65-78. - 1996. 'The Strange Disappearance of Civic America.' American Prospect 24 (Winter), www.epn.org/prospect/24/24putn.html. 1-19. Reich, Robert B. 1998. 'Broken Faith: Why We Need to Renew the Social Compact.' Nation (16 Feb.), 11-17. Ritzen, Jo William Easterly, and Michael Woolcock. 2000. 'On "Good" Politicians and "Bad Policies": Social Cohesion, Institutions, and Growth.' Seville, Spain: International Institute of Public Finance. Robison, Lindon J., and Marcelo E. Siles. 1997. 'Social Capital and Household Income Distributions in the United States: 1980, 1990.' Report 595. Michigan State University, Department of Agricultural Economics. Scott, FitzGerald. 1989. A New View of Economic Growth. Oxford: Clarendon Press. Stanley, Dick. 1997. 'The Economic Consequences of Social Cohesion.' Hull, Que.: Department of Heritage Canada, SRFA-302. Stiglitz, Joseph, and Jason Furman. 1999. 'Consequences of Rising Income Inequality.' Income Inequality: Issues and Policy Options. Kansas City: Federal Reserve Bank of Kansas City. Temple, Jonathon, and Paul A. Johnson. 1998. 'Social Capability and Economic Growth.' Quarterly Journal of Economics 113(3): 965-90.

230

M.C. McCracken

Weiner, Stuart E., and Stephen A. Montom. 1998. 'Income Inequality: A Summary of the Bank's 1998 Symposium.' Economic Review. Kansas City, Federal Reserve Bank of Kansas City, 1-11. Wilkinson, Richard G. 1996. Unhealthy Societies: The Afflictions of Inequality. London: Routledge. Woolcock, Michael. 1998. 'Social Capital and Economic Development: Towards a Theoretical Synthesis and Policy Framework.' Theory and Society 27(2): 151-208.

Many Happy Returns: How Social Cohesion Attracts Investment DICK STANLEY AND SANDRA SMELTZER

Traditional economics teaches that profit-maximizing firms invest when the expected present value of returns from an investment exceeds the expected present value of costs. However, in doing this, firms must take into account a series of more or less intangible factors that can affect expectations. The costs of doing business depend on issues such as the extent to which workers and partners can be trusted to respect contracts and laws, and hold up their ends of explicit and tacit bargains, the crime rate and costs of security, the degree of social dysfunction in society and the likelihood of it affecting the supply of reliable workers and the levels of productivity at work, the dependability of public services, the stability of the tax regimes, the trustworthiness or corruption of the political system, and the probability of social or political unrest. While low production costs occasioned by an abundance of natural resources and available land, low construction costs, low wages and statutory benefits, as well as the availability of reliable infrastructure and raw materials, low interest rates, and moderate taxes are conventionally thought to encourage investment (KPMG 1997). Traditional economics makes no mention of social cohesion as a variable in investment considerations. However, there are many ways in which a socially cohesive society has important economic advantages (Stanley 1997: 6). If abundant natural resources, low wages and benefits, and low taxes were the only factors that contribute to high levels of investment and economic growth, then the current economic status of Japan and Argentina would be quite reversed. As McRae (1994: 264) states, 'Countries can manage a burst of growth based on the exploitation of cheap labour and natural resources, and stomach a fair degree of corruption while they do so. But as the experience of a country like Argentina shows, it is hard to maintain prosperity amidst chaos.' Osberg (1992: 234) makes a similar point in stating, 'In the 1920s, Argentina was as rich a nation as Canada, and it remains a country

232 Dick Stanley and Sandra Smeltzer equally blessed with natural resources. Yet, political instability and economic stagnation have been characteristic of Argentinean life for the past 40 years. Although Japan's resource endowment is pitifully small compared to that of Canada, its social cohesiveness, from its opening to the outside world in the 1860s to the present day, has been legendary - and today Japan dominates many world financial and product markets.' Although by the end of the 1990s, the Japanese economy had lost some of its shine, Argentina's problems had become even worse. Japan remains one of the major world economies, and the comparison with Argentina remains dramatic. Alesina and Perotti (1993) repeat the same argument. Maddison (1995) suggests that to explain the differences in the economic performance of nations since the 1820s, one must look beyond natural resources, human capital, and the physical capital stock to what he calls 'ultimate causes.' The basic social order, as characterized by social institutions, beliefs, and ideology, as well as by the degree of sociopolitical conflict within a given social order 'can affect the rate of growth of the physical capital stock [investment], the quality of the labour force, the level and stability of demand, the rate of structural change, etc.' (1995: 88). In a similar vein, North (1990) argues that the institutions of a society determine to a significant degree the economic performance of nations. By institutions North means the complex of rules and arrangements through which citizens mediate their relationships with each other (e.g., the law and social norms). Institutions are developed by society to make up for imperfect information in social and economic transactions, to increase the predictability of outcomes, and thereby, to reduce transaction costs. Institutions can be dysfunctional to growth, if some societies reinforce traditional practices and make it irrational for economic actors to venture beyond traditional partnerships and trading arrangements. However, in other societies institutions provide safeguards and information to social and economic actors to give them confidence to reach beyond the length of their arms or their kinship groups and enter into the kind of complex and abstract partnerships necessary for advanced economic development. In other words, some societies have institutions that limit willingness to cooperate, and some societies have institutions that promote it. As these authors argue, factors other than the physical, financial, and fiscal must be taken into consideration when determining what makes up an attractive investment climate. This chapter argues that social cohesion plays an integral role in encouraging and shaping an attractive investment climate. In the first section, recent empirical evidence is presented to show that trust and willingness to cooperate partially explain economic growth and investment.

How Social Cohesion Attracts Investment

233

The second section argues that social cohesion reduces expected transaction costs through the provision of an amenity 'bonus,' through reducing the need for defensive necessities, and through the reduction in potentially costly political and labour instability. The third section emphasizes that social cohesion reduces expected costs by increasing worker productivity through the reduction in the incidence of individual social dysfunction and through the increased ease with which new ideas are disseminated. The fourth section discusses the evolution of the nature of work to suggest that social cohesion can be expected to be even more important in maintaining a favourable investment climate in the future. The conclusion suggests some of the implications for long-term government policy that arise from these arguments. Recent Empirical Evidence In this chapter, we build on our earlier work and define social cohesion as 'the bonding effect of society which arises spontaneously out of the unforced willingness of individuals to enter into relationships with each other and to contribute to the collectivity in order to enhance their survival, sustenance and success' (Stanley 1999). In other words, social cohesion arises because people are willing to cooperate with each other in mutual enterprises, to abide by the tacit or explicit rules of the relationships, and to contribute to relationships that do not bring them direct and immediate gain. One reason they do this may be because they recognize that, in the long term, such altruism will bring greater benefits to all, including themselves, than will selfish, shortterm actions. Recent empirical evidence in this vein includes Putnam (1993; 1996), Helliwell (with Putnam 1995), Alesina and Perotti (1993) and Knack and Keefer (1997). Helliwell and Putnam (1995) studied Italian regional governments over several decades. Although all regional governments were initially structured in a similar manner, the governments of some regions turned out to be efficient and effective, while others were disastrously inefficient and corrupt. Putnam and Helliwell discovered that the best empirical explanatory variable of the cause of these discrepancies was the strength of voluntary associational life - such as choral societies, football clubs, and other social groups and networks. The communities that had dense networks of civic engagement and a tradition of participation in these networks had better government. Moreover, civic connectedness was not only a crucial ingredient in explaining government efficiency, but also in explaining levels of economic well-being. Stable and well-functioning informal groups, such as choral societies, create social trust, reduce the costs of information and transactions, and decrease

234 Dick Stanley and Sandra Smeltzer economic risk. Therefore, they make individual actors more productive and make the area more attractive to investment, which explains the economic growth performance of these areas. Alesina and Perotti (1993) argue that, in addition to the traditional determinants of investment (such as relative costs of investment goods), levels of investment in a country are determined by the amount of human capital, the degree of equality in income distribution, and the degree of social and political stability. Alesina and Perotti maintain that income inequality fuels social discontent and increased social discontent fuels socio-political instability. This instability creates uncertainty in the political and economic environment, which reduces investment. These authors have created an index of sociopolitical instability (SPI) which is a weighted average of a number of variables such as attempted coups, deaths associated with mass violence, and degree of democracy. They argue that sociopolitical instability is a function of the levels of human capital, gross domestic product, investment, and income distribution. In turn, the level of investment is a function of sociopolitical stability, human capital, and the relative price of investment goods. Their results show that sociopolitical instability is an important factor in reducing investment, and that the greater the income inequality, the less investment. Alesina and Perotti (1993) also find that income inequality is highly correlated with the top income quintile getting rich at the expense of the next two quintiles (middle class). The investment-reducing effect of this unequal income distribution is stronger than any tendency of the richest quintile to save and so invest more. Finally, they find that countries with a higher education (more human capital) are more stable and therefore more attractive to investment. It also appears that sociopolitical instability depresses investment. The existence of a 'healthy' middle class (i.e., one which gets a reasonable share of the wealth of the economy and has the freedom to participate in civic life) is conducive to capital accumulation because it creates conditions of political stability. Differences in political stability between countries are not only statistically significant, but also economically significant. An increase in the political instability index of about one standard deviation causes a decrease of 4 per cent in the share of investment in gross domestic product (GDP). On this evidence, if Canada's level of political stability fell to that of France, our annual investment would be reduced by $10 billion dollars or roughly 5 per cent. Alesina and Perotti's findings suggest some relationship between investment and social cohesion, particularly through related concepts such as income equality and political stability. Knack and Keefer (1997) examine the direct effects of social cohesion on investment. Although they use the term

How Social Cohesion Attracts Investment 235 'social capital,' they are talking about the kinds of things this chapter has called social cohesion. Knack and Keefer's thesis is that there is a positive relationship between economic performance (by which they mean both economic growth and the investment to GDP ratio) and the degree to which interpersonal trust and adherence to norms of civic participation exist in the economy. As McCracken's chapter in this volume indicates, the implication of their estimates of the impact of level of social trust would be a substantial increase in growth and investment. They argue that trust lowers transaction costs, reduces expenses on such things such as security, and frees workers and entrepreneurs to concentrate on production and innovation rather than monitoring malfeasance. Because greater trust increases confidence in public policy, more investment is undertaken. Civic norms of cooperation act in somewhat the same way as trust, increasing productivity and reducing costs. Social cohesion was defined at the beginning of this section as the bonding effect of society which arises out of the willingness of people to cooperate with each other, based on trust and confidence. Moreover, social cohesion is based on the recognition that although contribution to the society as a whole may not bring immediate personal benefits, in the long run it will create significant benefits for the collectivity in which the individual will participate. In this way, social cohesion is critical to economic activity, since virtually all economic activity involves some form of human relationship or interaction. In the modern economy, where effectively no one operates completely independently, even the most 'Dickensian' factory organization involves some element of cooperation between labour and capital. The willingness to enter into and abide by the terms of relationships depends on trust or confidence in the response or reliability of partners. If trust is not present, creating the economic partnerships necessary for society as a whole to prosper becomes more difficult, if not impossible, and economic performance deteriorates. Even in our personal day-to-day economic dealings, trust matters. For example, if after-sales service matters, we try to avoid buying goods from an untrustworthy firm, regardless of the price. In fact, an unusually low price is often treated as a warning signal. Unless work effort can be exactly monitored (which is rare), we typically will not hire someone who strikes us as shiftless, unreliable, or sneaky, even though he or she may be willing to work for low wages, because we do not trust them to deliver a day's work for a day's pay (Stiglitz 1987). Although we know there is no such thing as an iron-clad guarantee that a business partner will deliver on her or his commitments (for countless possible causes, such as lightning striking the factory), we want to avoid the risk of opportunism or fraud. If we strike a deal with a

236 Dick Stanley and Sandra Smeltzer business partner, we recognize that there are possible risks, but we do not want to be taken advantage of. As Kenneth Arrow (1972: 357) submitted three decades ago, 'Virtually every commercial transaction has within itself an element of trust, certainly any transaction conducted over a period of time. It can plausibly be argued that much of the economic backwardness in the world can be explained by the lack of mutual confidence.' Similarly, Francis Fukuyama (1995: 27) suggests that 'people who do not trust one another will end up cooperating only under a system of formal rules and regulations, which have to be negotiated, agreed to, litigated, and enforced, sometimes by coercive means. Widespread distrust in a society, in other words, imposes a kind of tax on all forms of economic activity.' The existence of distrust in a society means that people accomplish less (i.e., their productivity is reduced) and requires people to go to greater lengths to accomplish even the most basic task (i.e., increases their transaction costs). These differentials in transaction costs can be an important source of regional comparative advantage. With falling costs of communication and transportation, and negotiated ends to trade barriers for many goods and services, even services whose production used to be tied to local markets can be produced elsewhere and sold in the local market. To attract investment, a country must have low unit costs, and transactions costs are part of this picture. Social Cohesion Reduces Transaction Costs Social cohesion reduces transaction costs in a number of ways: by increasing amenity values, by reducing the need for defensive necessities and risk in economic transactions, and by improving political and labour stability. Amenity Values By definition, social cohesion means people are willing to relate to each other within their community, lend a hand when someone else needs it, cooperate in mutual enterprises (like Putnam's associations), volunteer, and respect, uphold, and defend other people's rights. As Friesen's chapter in this volume discusses, such social cohesion can be of the 'bonding' (within small social groups) or 'bridging' (between social groups) variety. If social cohesion is primarily of a 'bridging' nature, a society tends to be caring, well functioning, and tolerant. In this type of society, the infrastructure works and personal security is high. There is respect for property and the

How Social Cohesion Attracts Investment 237 rights of the individual, and there is comparatively little corruption. As Mcpherson (1987: 2) notes, 'What distinguishes property from mere momentary possession is that property is a claim that will be enforced by society or the state, by custom or convention or law.' Claims to property are useless in a society where such claims are not respected. Property, which Mcpherson defines as the right in or to things, exists only by virtue of the general willingness of individuals to respect property rights and society's willingness to enforce individual rights when they are transgressed. In other words, without a shared consensus on the legitimacy of property rights (i.e., social cohesion) property as an institution could not exist. Corruption, then, or the unwillingness to recognize individual rights, is prima facie evidence of lack of social cohesion. As well, social cohesion has amenity value. The caring and tolerance of social life can be considered a sort of bonus on top of the pay cheque for individuals, which makes it cheaper, therefore, to attract labour and management to such a society. A number of business publications have made this point. As a Swedish diplomat has put it, 'the culture of a company is like the culture of a country. So Swedish businesses are accustomed to Canadian labour laws, to the notion of a social safety net and to environmental standards. With the high quality of life in Canada, it is very easy to lure executives to live here' (Evenson 1997). The accounting firm KPMG (1997: 8) recognized that quality of life factors are critical to a foreign investor. The KPMG report concluded that 'the final investment decision must go beyond cost factors, to include workforce availability, quality of life, education and medical care -just to name a few.' Defensive Necessities If social cohesion of the 'bridging' variety is to be maintained, members of society must demonstrate to each other trustworthiness and willingness to abide by society's rules and live up to their side of agreements. Since crime is an act of breaking an agreement, a socially cohesive society would be one where crime is the exception. Since violence is a way to force action from an unwilling partner when cooperation does not work, a socially cohesive society would be one where violence is similarly an exception. Greater 'bridging' social cohesion means less crime and violence, including theft, vandalism, and pilfering, both in public places and on the shopfloor. Hence, in a society that lacks such cohesion, a firm must take measures to protect itself, both legally (highly detailed contracts, defence against lawsuits, and insurance) and physically (security guards, locks and fences, and monitoring of employ-

238 Dick Stanley and Sandra Smeltzer ees). The senior executives who often accompany major investments also have to bear personal expenses in crime prevention - including, in some jurisdictions, the expense of bodyguards to prevent the possibility of kidnapping. In a cohesive society, these costs are lower and there is less need for formal legal protection, including complex contracts or the pursuit of legal action (see, e.g., Knack and Keefer 1997). In an important case study, Sampson et al. (1997) surveyed 8,782 residents of 343 Chicago neighbourhoods to measure the degree of social cohesion, which they called 'collective efficacy,' of each community and the degree of violent crime in each. The social cohesion of a neighbourhood was measured by the willingness of residents to undertake to a variety of activities that contribute to the collective good such as intervening to prevent truancy, confronting children engaged in minor vandalism, and intervening in various ways if the local fire station were facing budget cuts. The researchers also took into account attitudes of willingness to help neighbours and trust of neighbours. Violence was assessed both by residents' reports of their experiences with fights, violent arguments, assaults, and robberies, and by officially recorded incidents of homicide. Sampson et al. found that social cohesion was strongly negatively related to violence. In addition, when they examined more traditional factors thought to contribute to levels of violence, such as concentrations of economic disadvantage, immigrant concentration, and stability of residence, they found that it was the social cohesion of the neighbourhood that was mediating these more traditional contributing factors. Economic disadvantage and residential stability affected the levels of social cohesion, which in turn affected levels of violence. Where, for some reason or other residential stability, for example, did not increase social cohesion, it had much less tendency to lower violence. Thus, the cohesion that is created within these neighbourhoods helps to lower crime rates and create safe communities. In economic terms, this means that overall security costs are lower in socially cohesive communities because companies do not need to spend valuable resources on surveillance equipment or police or security guards to watch over and protect their businesses (McRae 1994: 16, 213; Osberg 1995: 14). Political and Labour Stability If a socially cohesive society means that individuals live up to their sides in agreements, it follows that social grievances tend to be solved through sanctioned channels, rather than outside a structure of rules (e.g., through violence). Even where there are conflicts, there could be expected to be respect

How Social Cohesion Attracts Investment 239 for property and human rights. As a result, there is less chance of loss of life and property, or of disruption of production processes, and a lower risk to investment. In addition, in a socially cohesive society, losers in policy changes may tend to believe that they will be compensated for their loss, and so tend to resist change less, lowering the cost of innovation (Osberg 1988; 1998). Hence, social cohesion lowers costs by reducing risk. As Alesina and Perotti (1993: 1) demonstrated, social discontent (i.e., a deterioration of social cohesion) fuels social unrest, which increases policy uncertainty and has a negative effect on investment, both domestic and foreign. As Knack and Keefer (1997: 1253) put it, 'Government officials in societies with higher trust may be perceived as more trustworthy, and their policy pronouncements as thus being more credible. To the extent that this is true, trust also triggers greater investment and other economic activity.' A similar case can be made for labour stability. If there is an initial belief among negotiating partners that they generally bargain in good faith (although they are each seeking to maximize their own benefits), then there is less tendency to resort to force or violence. This does not mean that, in a socially cohesive society, there would be no disagreements, because there will still be differences of interest (especially with respect to the sharing of profits) and differences in the information possessed (Herrington 1988), but in a community of shared norms and values the first recourse would be to resolve differences using institutionalized means. If such institutionalized means are believed in, used, and are successful (if they were not successful, they would soon cease to be believed in and used), the result is a more attractive (and cheaper) investment climate compared with places where labour disputes are resolved through use of violence and destruction of property. Social Cohesion and Productivity In addition to lowering transaction costs, social cohesion may increase the productivity of individual workers in three ways: (1) social cohesion decreases individual social dysfunction; (2) socially cohesive employees are happier and more satisfied with their work, and (3) social cohesion assists in the dissemination of new ideas. Social Dysfunction If social cohesion means that people are willing to relate to each other within their community, lend a hand when someone else needs it, and cooperate in

240 Dick Stanley and Sandra Smeltzer mutual enterprises (like Putnam's associations), volunteer, and uphold other people's rights, then individuals in cohesive societies are likely to have fewer personal problems (e.g., alcoholism, family break up, depression) or at least they would have the support to manage these problems better when they do have them. Since the likelihood of such individual social dysfunction is decreased, so also is absenteeism. The result is greater individual productivity (Osberg 1995: 14; Stanley 1997: 6). Bedard (1996: 11) and Osberg (1992) note that unemployment and economic insecurity, especially if it persists over an extended period, can lead to increased rates of social dysfunction (chronic depression, alcoholism, marriage break-up, family violence, and serious mental illness). The effects are felt not only by the individual, but also by his or her family, diminishing the life chances of the children. There is, therefore, an intergenerational compounding effect: children whose upbringing is free of social dysfunction are more likely to be successful in school and so more likely to be productive in later life (Phipps 1998; Statistics Canada 1998). Both unemployment and economic insecurity produce declines in productivity by making individuals less able to produce or in extremis unfit for the workforce. As Torjman (1997) has noted several Canadian business which had 'gone south in search of cheap labour and lower taxes returned to Canada. They found that there was something to be said for a healthy well-educated workforce and that Canada was, in effect, the "better bet.'" Dissemination of New Ideas If social networks are dense, and characterized by trust, people tend to talk to each other and to know how to judge what they hear. If, in a social context such as Putnam's choral society or bowling league, one hears of a new idea, a new technology, or a new way of organizing, the information comes embedded in a general knowledge of the person from whom it is learned. A choral society, for example, may create a great friendship with someone whom one knows to have a wonderful voice and no idea at all of electrical engineering. One may also get to know someone who is not very likable, but who is very knowledgeable about art history. Given this context, one has good reason not to waste time and energy on the first person's theory of computer design and to pay attention to the other person's recommendations on art purchases. It becomes easier, in short, to screen out bad ideas and select good ones. Thus, social cohesion opens up channels for the diffusion of new ideas and new technology and provides a lubricant to help ideas move more quickly from acquisition to acceptance. Similarly, in a cohesive workplace, where manage-

How Social Cohesion Attracts Investment 241 ment and labour have a tradition of cooperation and trust, both sides will be willing to float new ideas because they know that the organization will benefit and an improved organization will bring rewards to all. Stevens (1996) argues that the success of enterprises is determined by their effectiveness in gathering and using knowledge and technology, in other words, their ability to diffuse knowledge. The ability to diffuse knowledge depends on 'knowledge networks,' that is, the continuing interactions among institutions and individuals involved in the process of knowledge production and application: firms, laboratories, universities, and consumers. Mosco (1998: 8) also points to the importance of cities as 'vitally important centres for bringing together people, technology, institutions, finance and culture which feed on one another to attract new resources.' Jacobs (1970; 1984) has pointed out that the function of cities in bringing people together to share knowledge is critical to innovation and the effective diffusion of technology. As any researcher knows, it is crucial to know what did not work, but in general, published research talks about successful ideas. No published paper can report on all the details of how a technology has come to be workable, and by the time something is published, the research frontier has moved on. Networking enables the diffusion of early results, the discussion of unsuccessful research avenues, and the assessment of where research will go. Networking, therefore, is key to the diffusion of new technology. Networking and social cohesion are closely related because the informal information exchange between diverse individuals is based on trust and willingness to cooperate. As Thurow (1999) puts it, 'The right sociology [has] to be in place for revolutionary new products to emerge.' A socially cohesive workplace enables employees to bring forward tactical information which, while not necessarily innovative, will assist in improving decision making. For example, customer complaints about a product's operation will rise more quickly to decision levels in an organization where customer service representatives have confidence that problems will be solved rather than that blame will be attached (especially to them). This implies management will be better informed about production or marketing problems and will take action on better information sooner. Leo (1996) argues that, for example, Xerox Corporation almost went out of business in the early 1980s, owing to poor product quality. A concerted effort to empower employees, especially on the front line of customer service, to take corrective actions on their own and to send the knowledge of product improvement thus derived to management, and to make management more willing to act on and reward information offered by such employees, led to dramatic improvement in the company's fortunes.

242 Dick Stanley and Sandra Smeltzer The Evolution of Work and the Increasing Importance of Social Cohesion In the twenty-first century, to maintain productivity and competitiveness, a modern economy will increasingly require a more socially cohesive workforce. As the knowledge base and creative component of jobs grows, active intellectual participation and the sharing of knowledge and beliefs becomes increasingly important. As knowledge becomes an increasing proportion of the content of products and services, social cohesion increases in importance. Societies that have a tradition of social cohesion at all levels (e.g., the family, a small group, an institution, a nation) endow their members with an important job skill. The social skills that enable an individual to interact with his or her peers in the classroom, the schoolyard, and later in the community and the nation, turn out to be the same skills needed to prosper in the workforce of the new, knowledge-based economy. Osberg et al. (1995), in their survey of the quality and characteristics of the evolving labour market in Canada, illustrated through a series of case studies that the firms with the best chance of survival and with the greatest likelihood of producing 'good' jobs (jobs with high pay, security, career prospects, and reasonable working conditions) required a high component of social skills, that is, the willingness to cooperate in joint enterprises and to align one's individual goals to the goals of the collectivity. A number of authors have stressed the increasing need for social cohesion in the new information economy. Stewart (1997) argues that in the new economy and society, intellectual capital - not natural resources, machinery, or even financial capital - has become the one indispensable asset of corporations and countries. As Thurow (1992: 52), argues, 'In the coming century, the comparative advantage of nations will be entirely a human creation.' In this new competitive environment, communication skills and social ease - in essence the ability and the willingness to get along with co-workers - are what is necessary to encourage 'human creation' and increase productivity. In part, this shift is a result of a change in the output of firms. Carnevale and Rose (cited in Madrick 1998: 33) argue that there is a growing demand for people who work in offices, in which the majority are professionals, and a lesser proportion are support staff. Increasingly, such workers operate in a team environment - which puts an increasing premium on the ability to work with others. Although a traditional, assembly-line, manufacturing-based society must have skilled and knowledgeable workers, the success of the line is not dependent upon interpersonal relationships. However, a production pro-

How Social Cohesion Attracts Investment

243

cess such as described by Carnevale and Rose, which is based on creating knowledge, emphasizes interaction and cooperative intellectual activities. In this environment, socially cohesive behaviour becomes a necessity. Baldwin and Rafiquzzaman (1998) conclude that the trend in job creation in Canada is similar to that which Carnevale and Rose found in the United States. Conclusions There appears to be a strong relationship between the level of social cohesion in a society and the level of investment that it attracts. However, it is a subtle relationship with long lags. Thus, social cohesion can deteriorate for a lengthy period of time before this decline takes its economic toll on a society. Conversely, it can take a considerable length of time to build social cohesion back up again and before it starts to appear in the economic statistics. Looking at issues through the lense of social cohesion can give a new perspective on the costs and benefits for investment of a number of policy choices. The obvious conclusion of this essay is that economic policies that tend to reduce social cohesion (in the interests, e.g., of short-term fiscal deficit reduction) may, in fact, be counter-productive. If social stability unravels as cutbacks fuel discontent, the costs in lost productivity and increased transaction costs may well overshadow any assumed increase in the performance of the entrepreneurial class. As well, if the effects on social cohesion are considered, a policy of tax reduction intended to redistribute income to the wealthy (under the theory that they save and invest more) may, in fact, reduce investment. Short-run fiscal gain may be followed by long-run social pain. If the longer-term result of a less equitable tax burden is lower social cohesion, and if social cohesion affects (in a number of ways, as this essay has enumerated) the profitability of investment, one may ultimately end up with less investment and less growth. Although fiscal effects take place relatively quickly, it takes a greater amount of time for the effects of social cohesion to become apparent. When advocating public policy to protect or enhance our prosperity, we must examine a much broader and longer range of consequences than the narrowly economic. References Alesina, Alberto, and Roberto Perotti. 1993. Income Distribution, Political Instability, and Investment. Working Paper 4486. Cambridge, Mass.: National Bureau of Economic Research.

244 Dick Stanley and Sandra Smeltzer Arrow, Kenneth. 1972. 'Gifts and Exchanges.' Philosophy and Public Affairs 1: 343-62. Bedard, Marcel. 1996. The Economic and Social Costs of Unemployment. Ottawa: Applied Research Branch, Strategic Policy, Human Resources Development Canada (R-96-12E). Bohuslawsky, Maria. 1998. Texas Grass Only Looked Greener.' Ottawa Citizen, 25 July. Baldwin, John R., and Mohammed Rafiquzzaman. 1998. 'The Effect of Technology and Trade on Wage Differentials between Nonproduction and Production Workers in Canadian Manufacturing.' Research Paper Series, No. 98. Ottawa: Statistics Canada, Analytical Studies Branch. Boston Business Wire. 1998. 'Money Can't Buy Employee Commitment, WFD Research Reveals.' 4 Aug. Web site www.businesswire.com Evenson, Brad. 1997. 'Swedish Business Smitten with Canada: $1 billion Invested in the Past Year.' Ottawa Citizen, 2 July. Fukuyama, Francis. 1995. Trust: The Social Virtues and the Creation of Prosperity. New York: Free Press. Helliwell, J., with R. Putnam. 1995. 'Economic Growth and Social Capital in Italy'. Eastern Economic Journal 21(3): 295-307. Herrington, Douglas. 1988. 'The Effect of Private Information on Wage Settlement and Strike Activity.' Working Paper 231. Princeton, NJ: Princeton University Press. Industrial Relations Section. Jacobs, Jane. 1970. The Economy of Cities. New York: Vintage Books. - 1984. Cities and the Wealth of Nations. New York: Random House. Knack, Stephen, and Philip Keefer. 1997. 'Does Social Capital Have an Economic Payoff? A Cross-Country Investigation.' Quarterly Journal of Economics (Nov.): 1251-88. KPMG. 1997. The Competitive Alternative: A Comparison of Business Costs in Canada, Europe and the United States. (International Edition). KPMG Canada and Prospectus Inc. Leo, Richard J. 1996. 'Is TQM Dead? Not Having It Almost Killed Xerox.' 50th Annual Quality Congress Proceedings. Milwaukee: American Society for Quality, 485-95. Maddison, Angus. 1995. Explaining the Economic Performance of Nations. Aldershot, U.K.: Edward Elgar Publishing. Madrick, Jeff. 1998. 'Computers: Waiting for the Revolution.' New York Review 26 March, 29-33. Mcpherson, C.B., ed. 1987. Property: Mainstream and Critical Positions. Toronto: University of Toronto Press.

How Social Cohesion Attracts Investment 245 McRae, Hamish. 1994. The World in 2020: Power, Culture and Prosperity. Boston: Harvard Business School Press. Mosco, Vincent. 1998. 'The Socio-Cultural Implications of a Knowledge-Based Society: A Prospective Research Agenda.' Occasional Paper SRA-372. Ottawa: Strategic Research and Analysis Directorate, Department of Canadian Heritage. North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance. New York: Cambridge University Press. Osberg, Lars. 1988. 'Distributional Issues and the Future of the Welfare State.' K. Newton, T. Schweitzer, and J.-P. Voyer, eds., Perspective 2000: Proceedings of a Conference Sponsored by the Economic Council of Canada. Ottawa: Economic Council of Canada, 159-74. - 1992. 'Sustainable Social Development.' Robert C. Allen, and Gideon Rosenbluth, eds., False Promises: The Failure of Conservative Economics. Vancouver: New Star Books. 227-42. - 1995. 'The Equity / Efficiency Trade-off in Retrospect.' Canadian Business Economics (Spring): 5-19. - 1998. 'New Terms in the Economic Policy Debate.' Presentation for the Senate Standing Committee on Social Affairs, Science and Technology. 29 Oct., Ottawa. Osberg, Lars, Fred Wien, and Jan Crude. 1995. Vanishing Jobs: Canada's Changing Workplaces. Toronto: Lorimer. Phipps, Shelley. 1998. 'The Well-Being of Young Canadian Children in International Perspective.' Presented at the Conference on the State of Living Standards and the Quality of Life in Canada. Ottawa: 30-31 Oct. Putnam R. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. - 1996. 'The Decline in Civil Society: How Come? So What?' Optimum 27(1): 26-36. Sampson, R., S. Raudenbush, and F. Earls. 1997. 'Neighbourhoods and Violent Crime: A Multilevel Study of Collective Efficacy.' Science 111 (Aug.): 918-24. Stanley, Dick. 1997. 'The Economic Consequences of Social Cohesion.' Occasional Paper SRA-302. Ottawa: Strategic Research and Analysis Branch, Department of Canadian Heritage. Nov. - 1999. 'Hearing Secret Harmonies: Toward a Dynamic Model of Social Cohesion.' Ottawa: Strategic Research and Analysis Branch, Department of Canadian Heritage. Statistics Canada. 1998. 'National Longitudinal Survey of Children and Youth.' Daily, 28 Oct., 2-5. Stevens, Candice. 1996. 'The Knowledge-Driven Economy.' OECD Observer (June/July): 6-10.

246

Dick Stanley and Sandra Smeltzer

Stewart, Thomas A. 1997. Intellectual Capital: The New Wealth of Organizations. New York: Doubleday. Stiglitz, Joseph E. 1987. 'The Causes and Consequences of the Dependence of Quality on Price.' Journal of Economic Literature (March): 1-48. Swardson, Anne. 1995. 'Maple Leaf Manners.' Washington Post (13 Aug.), C01. Thurow, Lester C. 1992. Head to Head: The Coming Economic Battle among Japan, Europe, and America. New York: Morrow. - 1999. 'Building Wealth: The New Rules for Individuals, Companies, and Nations.' Atlantic Monthly 283(6): 57-69. Torjman, Sherri. 1997. 'On Magnets and Magnates.' Caledon Institute of Social Policy. May. www.caledoninst.org/full76.htm Yaffe, Barbara. 1998. 'Canada's Unique Values Protect us from "Brain Drain.'" Ottawa Citizen, 18 Nov., A9.

Contributors

Jeff Dayton-Johnson is an assistant professor of economics and international development studies at Dalhousie University. His research has focused on the economics of small-scale irrigation, cooperation, foreign aid, and culture and the arts, among other topics. He is the author of Social Cohesion and Economic Prosperity (2001). He received his PhD in economics from the University of California, Berkeley. Jane Friesen is associate professor of economics at Simon Fraser University. Following her undergraduate work in economics at the University of British Columbia, she received her PhD in economics from the University of Toronto. Her research interests include labour standards legislation, the distribution of hours of work, unemployment insurance, and labour demand. Her work has been published in a variety of academic journals in North America and Europe. John F. Helliwell is professor of economics at the University of British Columbia. His recent research has emphasized comparative macroeconomics and growth, including the influence of openness and institutions, as well as the importance of national borders. He is also involved in several interdisciplinary projects studying the linkages among economic, social and human health, and the role of social capital. He was Mackenzie King Visiting Professor of Canadian Studies at Harvard University from 1991 to 1994, and Christensen Visiting Fellow of St Catherine's College, Oxford, in 2001. He holds a DPhil. in economics from Oxford University. John N. Lavis holds an MD from Queen's University, an MSc from the London School of Economics, and a PhD in health policy from Harvard

248 Contributors University. He is the Canada Research Chair in Knowledge Transfer and Uptake at McMaster University and Liberty Health Scholar at The Canadian Institute for Advanced Research. He conducts research on knowledge transfer and uptake in public policy-making environments and on the social determinants of health. M.C. McCracken is chair and CEO of Infometerica Limited, an Ottawabased economic research and information company. The firm undertakes longterm economic forecasts on an industrial and regional basis, as well as economic consulting projects. His research includes the determinants of growth, rural economics, energy economics, and economic and social policy. Lars Osberg is McCulloch Professor of Economics at Dalhousie University. His major fields of research interest have been the determinants of poverty and economic inequality, with particular emphasis in recent years on social policy, and the implications of unemployment and structural change in labour markets for individuals and for social cohesion. As an undergraduate, he attended Queen's University, Kingston, and the London School of Economics and Political Science. He received his PhD in Economics from Yale University. Shelley Phipps is Maxwell Professor of Economics at Dalhousie University. Her current research focuses on the health and well being of Canadian children, international comparisons of social and family policy, poverty and inequality, and decision making within families. Her PhD is from the University of British Columbia. Gregory L. Stoddart, PhD, is a professor in the Centre for Health Economics and Policy Analysis, and the Department of Clinical Epidemiology and Biostatistics, and an associate member of the Department of Economics at McMaster University. He is also a Fellow in the Population Health Program of the Canadian Institute for Advanced Research. His published works cover topics ranging from the public-private mix in health care financing, methods for assessing cost-effectiveness of health services, and health human resource planning to social determinants of the health of populations, and cross-sectoral reallocation of resources for health. Dick Stanley is director of Strategic Research and Analysis for the Department of Canadian Heritage of the Government of Canada. He directs a team of social science researchers exploring issues of social cohesion, cultural

Contributors

249

diversity, and citizenship and identity. His current interests include the role of social cohesion in producing social well-being, and the effects of cultural diversity on social and community sustainability. He is a graduate of Carleton University and the New School for Social Research in Sociology. Sandra Smeltzer is currently working on her PhD in communications at Carleton University. As a research officer in the Strategic Research and Analysis Directorate of the Department of Canadian Heritage of the Government of Canada, she conducted extensive research in international comparative cultural policy and in the social and cultural impacts of new information technology. She holds an MA in anthropology from Carleton University. Frances Woolley is associate professor of economics at Carleton University. She has studied at Simon Eraser University and Queen's University, Kingston, and holds a doctorate from the London School of Economics. Her research focuses on applying economic theory to aspects of everyday life often neglected by economists, from voluntary activity and family decision making to crime and international development assistance.

This page intentionally left blank

Studies in Comparative Political Economy and Public Policy 1 The Search for Political Space: Globalization, Social Movements, and the Urban Political Experience I Warren Magnusson 2 Oil, the State, and Federalism: The Rise and Demise of Petro-Canada as a Statist Impulse I John Erik Possum 3 Defying Conventional Wisdom: Free Trade and the Rise of Popular Sector Politics in Canada I Jeffrey M. Ayres 4 Community, State, and Market on the North Atlantic Rim: Challenges to Modernity in the Fisheries I Richard Apostle, Gene Barrett, Peter Holm, Svein Jentoft, Leigh Mazany, Bonnie McCay, Knut H. Mikalsen 5 More with Less: Work Reorganization in the Canadian Mining Industry I Bob Russell 6 Visions for Privacy: Policy Approaches for the Digital Age I Edited by Colin J. Bennett and Rebecca Grant 7 New Democracies: Economic and Social Reform in Brazil, Chile, and Mexico I Michel Duquette 8 Poverty, Social Assistance, and the Employability of Mothers: Restructuring Welfare States I Maureen Baker and David Tippin 9 The Left's Dirty Job: The Politics of Industrial Restructuring in France and Spain IW. Rand Smith 10 Risky Business: Canada's Changing Science-Based Policy and Regulatory Regime I Edited by G. Bruce Doern and Ted Reed 11 Temporary Work: The Gendered Rise of a Precarious Employment Relationship I Leah Vosko 12 Who Cares? Women's Work, Childcare, and Welfare State Redesign I Jane Jenson and Mariette Sineau with Franca Bimbi, Anne-Marie DauneRichard, Vincent Delia Sala, Rianne Mahon, Berenger Marques-Pereira, Olivier Paye, and George Ross 13 Canadian Forest Policy: Adapting to Change I Edited by Michael Howlett 14 Knowledge and Economic Conduct: The Social Foundations of the Modern Economy I Nico Stehr 15 Contingent Work, Disrupted Lives: Labour and Community in the New Rural Economy I Anthony Winson and Belinda Leach 16 The Economic Implications of Social Cohesion I Edited by Lars Osberg