social capital in solid waste management: evidence from dhaka

Social Capital Initiative Working Paper No. 16

SOCIAL CAPITAL IN SOLID WASTE MANAGEMENT: EVIDENCE FROM DHAKA, BANGLADESH

Sheoli Pargal, Mainul Huq, and Daniel Gilligan The World Bank Social Development Family Environmentally and Socially Sustainable Development Network September 1999

ii

Working papers can be viewed at http://www.worldbank.org/socialdevelopment, or obtained from:

The World Bank Social Development Department Social Capital Working Paper Series Attention Ms. Gracie M. Ochieng 1818 H Street, NW, Room MC 5-410 Washington, DC 20433, USA

Tel: (202) 473-1123 Fax: (202) 522-3247 Email: [email protected]

or: Social Development Department The World Bank 1818 H Street, NW, Room MC 5-232 Washington, DC 20433, USA

Fax: (202) 522-3247 Email: [email protected]

Papers in the Social Capital Initiative Working Paper Series are not formal publications of the World Bank. They are published informally and circulated to encourage discussion and comment within the development community. The findings, interpretations, judgements, and conclusions expressed in this paper are those of the author(s) and should not be attributed to the World Bank, to its affiliated organizations, or to members of the Board of Executive Directors or the governments they represent.

i

SOCIAL CAPITAL INITIATIVE WORKING PAPER SERIES #1 The Initiative on Defining, Monitoring and Measuring Social Capital: Overview and Program Description

#2 The Initiative on Defining, Monitoring and Measuring Social Capital: Text of Proposals Approved for Funding

#3 Social Capital: The Missing Link? (by Christiaan Grootaert)

#4 Social Capital and Poverty (by Paul Collier)

#5 Social Capital: Conceptual Frameworks and Empirical Evidence An Annotated Bibliography (by Tine Rossing Feldman and Susan Assaf)

#6 Getting Things Done in an Anti-Modern Society: Social Capital Networks in Russia (by Richard Rose)

#7 Social Capital, Growth and Poverty: A Survey and Extensions (by Stephen Knack)

#8 Does Social Capital Facilitate the Poor’s Access to Credit? A Review of the Microeconomic Literature (by Thierry van Bastelaer)

#9 Does Social Capital Matter in Water and Sanitation Delivery? A Review of Literature (by Satu Kähkönen)

#10 Social Capital and Rural Development: A Discussion of Issues (by Casper Sorensen)

#11 Is Social Capital an Effective Smoke Condenser?: An Essay on a Concept Linking the Social Sciences (by Martin Paldam and Gert Tinggaard Svendsen)

#12 Ethnicity, Capital Formation, and Conflict (by Robert Bates)

#13 Mapping and Measuring Social Capital: A Conceptual and Empirical Study of Collective Action for Conserving and Developing Watersheds in Rajasthan, India (by Anirudh Krishna and Norman Uphoff)

#14 What Determines the Effectiveness of Community-Based Water Projects? Evidence from Central Java, Indonesia on Demand Responsiveness, Service Rules, and Social Capital (by Jonathan Isham and Satu Kähkönen)

#15 What Does Social Capital Add to Individual Welfare (by Richard Rose)

#16 Social Capital in Solid Waste Management: Evidence from Dhaka, Bangladesh (by Sheoli Pargal, Mainul Huq, and Daniel Gilligan)

#17 Social Capital and the Firm: Evidence from Agricultural Trade (by Marcel Fafchamps and Bart Minten)

#18 Exploring the Concept of Social Capital and its Relevance for Community-based Development: The Case of Coal Mining Areas in Orissa, India (by Enrique Pantoja)

#19 Induced Social Capital and Federations of the Rural Poor (by Anthony Bebbington and Thomas Carroll)

#20 Does Development Assistance Help Build Social Capital? (by Mary Kay Gugerty and Michael Kremer)

#21 Cross-cultural Measures of Social Capital: A Tool and Results from India and Panama (by Anirudh Krishna and Elizabeth Shrader)

#22 Understanding Social Capital. Agricultural Extension in Mali: Trust and Social Cohesion (by Catherine Reid and Lawrence Salmen)

#23 The Nexus between Violent Conflict, Social Capital and Social Cohesion: Case Studies from Cambodia and Rwanda (by Nat J. Colletta and Michelle L. Cullen)

#24 Understanding and Measuring Social Capital: A Synthesis of Findings and Recommendation from the Social Capital Initiative (by Christiaan Grootaert and Thierry van Bastelaer)

iii

FOREWORD There is growing empirical evidence that social capital contributes significantly to sustainable development. Sustainability is to leave future generations as many, or more, opportunities as we ourselves have had. Growing opportunity requires an expanding stock of capital. The traditional composition of natural capital, physical or produced capital, and human capital needs to be broadened to include social capital. Social capital refers to the internal social and cultural coherence of society, the norms and values that govern interactions among people and the institutions in which they are embedded. Social capital is the glue that holds societies together and without which there can be no economic growth or human well-being. Without social capital, society at large will collapse, and today’s world presents some very sad examples of this. The challenge of development agencies such as the World Bank is to operationalize the concept of social capital and to demonstrate how and how much it affects development outcomes. Ways need to be found to create an environment supportive of the emergence of social capital as well as to invest in it directly. These are the objectives of the Social Capital Initiative (SCI). With the help of a generous grant of the Government of Denmark, the Initiative has funded a set of twelve projects which will help define and measure social capital in better ways, and lead to improved monitoring of the stock, evolution and impact of social capital. The SCI seeks to provide empirical evidence from more than a dozen countries, as a basis to design better development interventions which can both safeguard existing social capital and promote the creation of new social capital. This working paper series reports on the progress of the SCI. It hopes to contribute to the international debate on the role of social capital as an element of sustainable development.

Ismail Serageldin Vice-President

Special Programs

v

THE INITIATIVE ON DEFINING, MONITORING AND MEASURING SOCIAL CAPITAL

STEERING COMMITTEE Ismail Serageldin (Vice-President, Special Programs) Gloria Davis (Director, Social Development Department) John Dixon (Chief, Indicators and Environmental Valuation Unit) Gregory Ingram (Administrator, Research Advisory Staff) Emmanuel Jimenez (Research Manager, Development Economics) Steen Lau Jorgensen (Sector Manager, Social Protection, Human Development Department) Peter Nannestad (Professor of Political Science, University of Aarhus, Denmark) John O’Connor (Consultant) Charles Cadwell (Principal Investigator, IRIS Center, University of Maryland) Martin Paldam (Professor, Department of Economics, University of Aarhus, Denmark) Robert Picciotto (Director General, Operations Evaluation) Gert Svendsen (Assistant Professor of Economics, Aarhus Business School, Denmark)

STAFF Christiaan Grootaert (Task Manager) Thierry van Bastelaer (Coordinator) Susan Assaf (Consultant) Casper Sorensen (Consultant) Gracie Ochieng (Program Assistant)

vii


ABOUT THE AUTHORS

Sheoli Pargal, currently in the Private Sector Development group of the World Bank’s Latin America region (LAC), works on infrastructure regulation and improving the business environment in LAC countries.1 Mainul Huq is currently working as an environmental economist in the Development Policy Group (DPG), Bangladesh. Formerly, he worked in the World Bank at DECRG Environment and Infrastructure. His primary research area is industrial pollution policy. One of his main interests is research on how social capital aids technology diffusion, pollution monitoring, and environmental improvement. Dan Gilligan is a graduate student in the program of Agricultural Economics at the University of Maryland.

ACKNOWLEDGMENTS We thank Thierry van Bastelaer and Christiaan Grootaert of the Social Capital Initiative for their help and comments and thank Jyotsna Jalan for very useful suggestions.

1 Please address comments to [email protected]

ix


TABLE OF CONTENTS

1. INTRODUCTION..............................................................................................................1 2. SOCIAL CAPITAL ...........................................................................................................2 3. MODELING THE DEVELOPMENT OF VSWM SYSTEMS...............................................5 4. ESTIMATION STRATEGY ...............................................................................................8 5. DATA COLLECTION.....................................................................................................10 6. DATA DESCRIPTION ....................................................................................................12 7. RESULTS.......................................................................................................................16 8. CONCLUSIONS..............................................................................................................21 REFERENCES .....................................................................................................................22 APPENDIX ..........................................................................................................................26

1

1. INTRODUCTION This paper seeks to identify the role played by social capital in the private, community-based provision of a public good, in this case, trash collection. The community aspect is vitally important here since trash collection involves positive externalities leading to limited incentives for individual action. Also, trash collection is an activity in which individual action does not have much impact, so collective action is warranted. Why are some communities better able to organize themselves for the collective good than others? Given the same impetus, what particular characteristics of the community lead to activism in some neighborhoods and none in others?

The context for this work is the fact that households in some neighborhoods of Dhaka, Bangladesh, have organized themselves to arrange for private collection of trash. The garbage collection system in Dhaka involves municipal pick-up from large dumpsters placed in central areas, with municipal workers responsible for collecting trash from smaller dumpsters located in alleys and side streets and transporting it to the main dumpsters. However, municipal employees are unreliable and frequently fail to collect the trash on a regular basis. In response, some communities have hired private contractors to undertake local trash collection funded by voluntary contributions from community members. Since other, apparently similar, neighborhoods have not managed to successfully organize an alternative to the municipal service, a natural question is why some communities or neighborhoods display such initiative while others do not.

We conjecture that “social capital” is a critical determinant of such collective action, where we equate social capital with community cohesiveness. The cohesiveness of the community is, in turn, a function of factors like customary or traditional interactions and institutions, a common heritage, values, ethnic or religious background, etc. Using data obtained from a survey of neighborhoods in Dhaka, we examine the importance of these potential determinants of social capital in the establishment of voluntary solid waste management (VSWM) systems as exemplified by the existence of trash disposal committees in these neighborhoods. We view the creation of these committees as a direct benefit of collective action, which is a function of the social capital in the neighborhood.

Using measures of trust and the strength of norms of reciprocity and sharing among neighborhood residents as proxies for social capital, we find that social capital is, indeed, an important determinant of whether VSWM systems arise in Dhaka. The effects of norms of reciprocity and sharing on the probability that a VSWM system is created are relatively large and significant, while the role of trust is not identified as a significant factor. Other measures of homogeneity of interests are also important, and, interestingly, so is the nature of associational activity. Finally, as would be expected, education levels are strongly and robustly associated with the existence of collective action for trash disposal.

The remainder of this paper is structured as follows. In section 2, we briefly describe the literature on social capital relevant to this study. Following that, we describe our modeling and empirical estimation strategy in sections 3 and 4, and our survey and data in sections 5 and 6. We present our empirical results and discuss their implications in section 7. Section 8 concludes.

2

2. SOCIAL CAPITAL The term 'social capital' has been applied to a variety of ideas that generally concern economic returns from networks of social relationships. While there has been limited work in economics on providing a theoretical context for social capital, there is a growing empirical literature that identifies considerable economic returns to networks of social relationships, to trust and norms of reciprocity, and to institutions that foster civic engagement.

Social capital first gained popularity and analytical teeth from James S. Coleman's works (1988, 1990). Citing Loury's (1977) definition of social capital as “the set of resources that inhere in family relations and in community social organization”… (1990, p. 300), Coleman sees social capital as the "social relationships which come into existence when individuals attempt to make best use of their individual resources" (1990, p. 300).

Like other forms of capital, social capital is productive, making possible the achievement of certain ends that would not be attainable in its absence. Like physical capital and human capital, social capital is not completely fungible, but is fungible with respect to specific activities.... Unlike other forms of capital, 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, p. 302)

While Coleman stresses social capital as resources that accrue to individuals, Putnam (1993) popularized a definition of social capital as resources that can characterize societies: "Social capital here refers to features of social organization, such as trust, norms, and networks, that can improve the efficiency of society by facilitating coordinated actions" (1993, p. 167). Putnam is concerned not only with the role of social capital in economic development, but also with its role in forming democratic societies. Thus, he equates social capital with intensity of 'civic engagement.'2

Social networks can be characterized as primarily ‘horizontal’, in which individuals share relatively equal status and power, or primarily ‘vertical,’ with asymmetric relationships based on hierarchy and dependence. Putnam argues that horizontal networks such as “neighborhood associations, choral societies, cooperatives, sports clubs, mass-based parties, and the like” (1993, p. 173) are the building blocks of ‘networks of civic engagement.’ These networks are “an essential form of social capital: The denser such networks in a community, the more likely that

2 There are a number of studies – both theoretical and empirical – that consider the economic benefits of trust (Fukuyama (1995), Narayan and Pritchett (1997), Knack and Keefer (1997)), norms of reciprocity (Sugden (1984, 1986)), culture (Harrison (1992), Greif (1994)), and ethnicity (Borjas (1992, 1995)) either as (sometimes implicit) determinants of or proxies for social capital. Putnam argued that the quality and intensity of individual membership in social and professional associations is a good indicator of social capital. This indicator of social capital has been used by many researchers to test the benefits of social capital (see Narayan and Pritchett (1997), Knack and Keefer (1997), Helliwell and Putnam (1995), Meyerson (1994), and Boxman, De Graaf and Flap (1991), for example).

3

its citizens will be able to cooperate for mutual benefit” (1993, p. 173).3 Trust and reciprocity to sustain civic networks (i.e., social capital) are self-reinforcing because as these networks become more dense the costs of opportunistic or selfish behavior increase. This implies endogeneity in regressions that attempt to explain a product of social capital with some measure of trust or reciprocity, for instance.

There are several mechanisms through which social capital might affect economic outcomes. Repeated interaction by economic actors through social networks strengthens trust, lowers transaction costs, improves the flow of information (reduces information asymmetries) and increases the enforceability of contracts. As a result, social capital can increase welfare by increasing the likelihood of cooperative behavior in prisoner's-dilemma-type problems, in the private provision of public goods, and in the management of common property resources (Ostrom (1990, 1996), Baland and Platteau (1996), White and Runge (1994)).

One of the first rigorous empirical studies of social capital is Narayan and Pritchett (1997). Following Putnam, they measure social capital using involvement in civic and professional associations and show that at both the household and village level, social capital is a significant determinant of income for a sample of households in Tanzania.4 In order to remove the potential endogeneity of social capital due to simultaneous effects of income on associational activity (which would result if social capital were a consumption good), the authors instrument for social capital using indicators of trust from survey questions.5

Knack and Keefer (1997) in a cross-country empirical study, show that social capital matters for economic growth, using indicators of trust and of civic cooperation as direct measures of social capital. They deal with the potential endogeneity of social capital in their regressions of economic performance by using performance data that is subsequent to the measures of trust and civic cooperation. Testing the importance of Putnam's horizontal networks by measuring the effect of associational activity on trust, civic cooperation, and economic growth, they find no relationship between associational life and these measures.

There has been little work on the role of social capital in solid waste management (SWM). An insightful article by Beall (1996) presents case studies of cooperative action for the provision of solid waste management. In Bangalore, India, Beall found free riding and caste considerations undermined the effort of horizontal associations of NGOs to organize

3 While Coleman offers a definition of social capital as a capital asset for the individual, Putnam treats the aggregation of social networks as representing the social capital of a society. In a critique of Putnam, Harriss and De Renzio (1997) question whether this type of ‘scaling up’ is consistent with the idea of the strength of social ties as a form of capital. In particular, because social organization can also be used to exclude others from economic benefits or for rent-seeking as noted by Olson (1982), the aggregate gain from summing individuals’ benefits from social interaction may be ambiguous. 4 Household social capital is measured by the number of associations to which respondents belong, with membership weighted by the quality of the association in raising social capital. The latter is measured by characteristics of the associations and the respondents’ assessment of the trust and social cohesion in the group (including kin heterogeneity, income heterogeneity, group functioning, group decision making, and voluntary membership). Village-level social capital is defined as the product of the average number of groups per household times the average group characteristics. 5 This is consistent with Fukuyama's (1995) notion that trust, in part, determines the effectiveness of social capital.

4

neighborhood based SWM. In Faisalabad, Pakistan, Beall found that richer neighborhoods often had sufficient political clout to guarantee continued government services while much of the rest of the city went without. Poorer communities were sometimes able to gain access to public services by offering their neighborhoods as vote banks, guaranteeing support at the polls in exchange for electricity, sewerage and the like. In Pakistan, it was vertical rather than horizontal networks between neighborhood leaders and politicians that enabled the provision of public goods by the government.6

6 An important insight of Beall’s work is that, in the rush to decentralize government, many have turned to social capital as a potential mechanism for the private provision of public goods. Attempts to tack additional services, such as security details or health committees, onto existing local cooperative efforts may fail – because they ignore the characteristics of the problem that originally caused people to organize.

5

3. MODELING THE DEVELOPMENT OF VSWM SYSTEMS We use a threshold model of public good provision, following Black, Levi, and de Meza (1993). We assume that initiators propose the creation of a VSWM system to neighborhood residents for their consideration. The initiators will form the neighborhood VSWM system only if a large enough number of households decide to participate.7 Individual households undertake a cost-benefit analysis to decide whether to join the proposed system, recognizing the impact of their decision on the probability of the system coming into being. Our field work surrounding data collection indicates that this analytical context for VSWM system formation is a good representation of the actual process occurring in the neighborhoods of Dhaka.

Let N be the number of households in the neighborhood, and n the threshold number of participating households necessary for VSWM system formation. Let the cost per household of participation, c, be declining in the level of neighborhood social capital ( )( )0<′ kc due to increased ease of coordination and easier flow of knowledge as social capital increases. The private benefits to household i of trash disposal, bi, are augmented by the household’s social capital, ki, so that total private benefits from joining the VSWM system are ( )ii bkB , .8 The effect of household social capital on total private benefits from joining the VSWM system depends on the nature of social norms in the neighborhood. If norms of reciprocity are strong, we expect that households with stronger ties in the community (higher ki) earn greater (net) rewards for their cooperation by reinforcing their standing and from the act of participating in a community initiative ( )( )0,1 >ii bkB .9 We can assume that these benefits are additive in ki and bi, so that if, for example, reciprocity norms are well developed households with strong ties in the neighborhood but low concern for public cleanliness may earn as much direct benefit from joining the system as a household of environmentally conscious members that does not have close neighborhood ties. We also assume that ( ) ( )kcbkB ii <, for all i = 1,…, N so that no household is willing to act alone.

In addition, the benefits of trash removal as a public good are assumed to be linear in the number of households that enroll in the system, so that a household receives an additional benefit equal to bi for each other household that joins up. Thus, if r other households join, the ith household’s benefits are augmented by bir. The probability that r other households will join the system is given by Pr(r| N-1, k), which is a function of the number of other households in the

7 In order for the proposed project to be viable, a minimum number of households is required in order to cover fixed costs associated with operating the system (such as the purchase of a cart to transport the trash). 8 Here, household social capital, ki, is restricted to the social capital that exists in the household’s relationships with other households in its neighborhood, whereas neighborhood social capital, k, is a measure of the average social capital of all households in the neighborhood. 9 However, more dynamic norms of community engagement might imply that households with weaker social ties in the community are rewarded more generously for their cooperation as a method of encouraging greater future community participation ( )( )0,1 <ii bkB .

6

neighborhood, N-1, and k, the level of neighborhood social capital. The VSWM system will only come into being if r>n.10

Households are assumed to act non-strategically, taking the decisions of other households as given (Cournot-Nash behavior). Thus the ith household will commit to participating if the expected benefits from taking part exceed the expected costs:

(1) ( ) ( )[ ] ( ) ( )∑∑∑−

=

−

−=

−

−=

−>−−+−11

1

1

1

,1|Pr,1|Pr)(,,1|PrN

nri

N

nrii

N

nri kNrrbkNrkcbkBkNrrb .

The first term above represents the expected benefits to the ith household of at least n-1 other households agreeing to join the VSWM system when the ith household joins, so that the system is formed and trash is collected (the public good element). The second term represents the direct expected net benefits to the ith household of its own contribution. The term to the right of the inequality is the expected (non-excludable) benefits to the ith household assuming that it decides not to join but the VSWM system is formed anyway.

This inequality captures the tradeoff faced by each household in its decision to join. The household prefers that a VSWM system be created but that it avoid the costs of joining. However, by free-riding the household also runs the risk of being the critical vote that keeps the neighborhood from achieving the n members required.11

The probability of a VSWM system being formed in neighborhood j is the probability that at least n households agree to participate. Let r~ be the number of households that agree to join. Defining yj =1 if a VSWM system is formed in neighborhood j, and yj =0 otherwise,

(2) ( ) ( ) ( ) ( )( )∑=

=>==N

nrNNj kpkpNrnry

~11 ,...,,|~Pr~Pr1Pr ,

where ( )ii kp is the probability of the ith household joining the VSWM system, i.e., the probability that the participation constraint in (1) is satisfied for the ith household:

(3) ( ) ( ) ( )[ ] ( ) ( )

−>−−+−= ∑∑∑

−

=

−

−=

−

−=

11

1

1

1

,1|Pr,1|Pr)(,,1|PrPrN

nri

N

nrii

N

nriii kNrrbkNrkcbkBkNrrbkp

This suggests a probit model of neighborhood cooperation in which the latent variable, ∗jy ,

measures the intensity of cooperation for public good provision in the jth neighborhood. In the next section, we describe the empirical implementation of this model.

Obviously household benefits from the VSWM systems are not observed and the costs are known only in neighborhoods where the system exists. However, under the assumption of 10 As households do not observe the threshold for the existence of a VSWM system, we assume it is exogeneous. 11 Many of the TDCs in Dhaka provide two services, collection of trash from household bins and removal from centralized neighborhood bins. The latter service is clearly a neighborhood public good. Although trash collection at the household is a private good, when the trash is not collected regularly it often ends up in public areas such as streets and vacant lots. The model developed here focuses on the public good component of these services, assuming that private gains from household trash removal are already optimally provided.

7

Cournot-Nash behavior in which household j does not take account of the decision process of household i, an increase in ( )ii kp for an arbitrary household increases the probability of VSWM system formation in (2). Thus, (2) and (3) can be used to identify how a number of other variables affect the probability of such systems coming into being.

For example, if information costs are increasing in the number of households in the neighborhood or if coordinated action is simply more difficult as the number of actors increases, then the probability of VSWM system formation will be decreasing in the number of households in the neighborhood. On the other hand, a larger neighborhood implies a larger potential set of participants, so that a given threshold, n, may be easier to reach in larger neighborhoods. Education should increase the probability of cooperation by increasing the perceived benefits from environmental improvements. The degree of homogeneity of ethnic or regional origin of neighborhood residents may also affect cooperation, although here again there are arguments in favor of both positive and negative impacts. One might expect that people from the same ethnic group or hometown will have stronger social ties and be more likely to work together. However, these same ties can reinforce inertia and a reluctance to change old ways of doing things. Income may have little direct effect on the probability of cooperation, but it may proxy for other variables that can affect the costs and benefits of joining a VSWM system. At higher income levels, the actual cost of joining a VSWM system may represent a very small share of total expenditure suggesting that it would be easier for these households to join. However, we suspect that higher income neighborhoods have better municipal trash removal services due to greater influence with local politicians (a form of social capital not captured in our survey), so that the benefits from a VSWM system will be lower in these neighborhoods.12 Variables for each of these determinants of cooperation are included in the neighborhood probit below.

From (2) and (3), we can also determine the probable effect of social capital on the probability that a VSWM system will form. Neighborhood social capital reduces the costs

( )kc of participating in collective activity, thereby increasing the probability of greater cooperation. In addition, household social capital may increase the direct benefits ( )ii bkB , of participation if norms of reciprocity are strong.13 The main hypothesis in this study thus is that higher levels of neighborhood social capital increase the probability of the formation of trash disposal committees. An intermediate hypothesis is that greater neighborhood social capital increases the household’s perceived probability that r other households will agree to participate in the system ( )[ ]( )0,1|Pr >∂−∂ kkNr .

12 Unfortunately we have no indicators of the quality of municipal trash disposal services prior to the initiation of VSWM systems, so we are unable to control for the effect of this factor. 13 Norms of reciprocity are quite strong in the neighborhoods of Dhaka included in our sample, as shown in the discussion of the data below. As a result, we expect any measure of social capital that incorporates norms of reciprocity to have a positive effect on the probability of VSWM formation through its effect on the benefits of cooperation ( 0),(1 >ii bkB ).

8

4. ESTIMATION STRATEGY As mentioned in the discussion above we decided to implement a probit model of neighborhood cooperation in which the latent variable, ∗

iy , measures the intensity of cooperation for public good provision in the ith neighborhood. Cooperation is a function of the level of neighborhood social capital, the number of households, income and education levels, and ethnic homogeneity. However, there is a simultaneous relationship between social capital and the establishment of a VSWM system since the process of developing a VSWM system can strengthen social networks, may encourage participation in other civic and social organizations, and can build trust and reinforce norms of reciprocity. As a result, any contemporaneous measures of social capital used as regressors in the cooperation equation will be correlated with the error term. To account for this source of endogeneity, we employ a two-stage conditional maximum likelihood (2SCML) estimation procedure developed by Rivers and Vuong (1988).

We estimate a discrete choice simultaneous equations system which includes three equations that determine contemporaneous continuous measures of social capital (here trust, reciprocity and sharing) plus a probit regression for cooperation with endogenous social capital. The ith observation in the system is represented as:

iiii

iii

uXYy

VXY

+′+′=

+Π′=∗ βγ 1 )b.4(

)a.4(

where i=1,…,n, and iY , iX 1 , and iX , are 1×m , 1×k , and 1×p vectors, respectively.14 A * denotes a latent variable. Equations (4.a) include m social capital regressions written in reduced form with observations on n neighborhoods. Equation (4.b) is a probit regression to explain latent neighborhood cooperation for public good provision. Although cooperation is unobserved, we observe whether or not a VSWM system exists, which we represent by the binary variable, iy , defined as

otherwise. 0

0 if 1

=>= ∗

i

ii

y

yy

In order to obtain consistent parameter estimates, rewrite equation (4.b) in the form

iiiii VXYy ηλβγ +′+′+′=∗1 )5( , where λη iii Vu ′−= . Rivers and Vuong (1988) show that an

appropriate normalization for this system is (6) 1=Σ′− λλσ vvuu , where vvΣ is the covariance matrix of residuals from the social capital regressions. Estimation of the conditional ML probit regression then takes place in two steps. First, estimate the social capital regressions

14 The matrices in the social capital regressions and the cooperation probit are related by the identity ii XJX ′=1 ,where J is a selection matrix of zeros and ones that retrieves iX 1 from iX .

9

to obtain Π̂ and vvΣ̂ , whose diagonal elements are estimated by i

n

iiVVn ′∑

=

ˆˆ11

where

iii XYV Π′−= ˆˆ . Then, estimate the probit regression for (5) substituting iV̂ for iV to obtain

)ˆ,ˆ,ˆ( λβγ . Greater detail on the estimation procedure is provided in Appendix B.

Hausman tests for exogeneity did not allow us to reject (at 5% significance) the exogeneity of the existence of VSWM systems in the social capital equations. Since we are interested in obtaining efficient estimates of the effects of cooperation on social capital, we estimate a modified version of equations (4.a) including the dummy variable for the presence of a VSWM system as a regressor as well as only the subset of exogenous variables from X that we believe are determinants of our contemporaneous measures of social capital. These include measures of previous neighborhood participation in civic associations (past social capital), tenure of neighborhood residents, regional origin of neighborhood residents, employment status, the share of residents that own their home, and the availability of infrastructure to facilitate community meetings. Under the (strong) assumption of spherical disturbances in the social capital regressions with no cross equation correlation (i.e., that vvΣ̂ is diagonal) maximum likelihood estimates of the social capital regressions can be obtained by least squares estimation.

We use the Rivers and Vuong estimation method because their Monte Carlo results demonstrate that their estimator performs better than the alternative Amemiya (1978) estimator in small samples, and our sample from Dhaka covers only 65 neighborhoods. Another advantage of the Rivers and Vuong procedure is that it is possible to develop a simple Wald statistic from parameter estimates for λ to test for exogeneity of the social capital variables in the cooperation probit.

10

5. DATA COLLECTION We undertook a survey of voluntary solid waste management practices in Dhaka between November 1997 and January 1998 using a structured questionnaire and interviewing households in sixty-five lower to upper middle-class neighborhoods of the city. The object of the survey was to collect household and neighborhood level information that would allow us to construct measures of associational activity, trust, reciprocity and sharing, as well as learn about neighborhood characteristics that might explain the establishment of the VSWM systems that exist in some of these neighborhoods.

The survey was confined to Dhaka City proper. Solid waste management in Dhaka comes under the purview of the Dhaka City Corporation (DCC), which covers approximately 360 square kilometers. Rich neighborhoods were not in the sample, as the corporation-run system seems to work well in such areas. The poorest neighborhoods (slum areas) were not included either since no VSWM systems exist there.

The structured questionnaire had three modules. The first module dealt with general information regarding the community such as the number of residents, their districts of origin, mix of landlords and tenants, age of the neighborhood, etc. It also recorded the number of associations, frequency of meeting, membership, etc. The second module related to data about the households that participated in the survey. Questions on income/expenditure, education, age and profession of the members of the household were asked about in this section, as well as questions regarding trust, reciprocity, and sharing, that formed the bases of our proxies for social capital. The final module pertained to the initiators of the VSWM systems in areas which had functioning systems, in an attempt to understand the motivation and characteristics of VSWM initiators.

The implementation of the survey consisted of three distinct phases: scouting, pre-testing of the questionnaire, and interviewing heads of households, and the initiators. Scouting consisted of identifying the neighborhoods where such systems exist, as there was no list of such neighborhoods available. The population of neighborhoods with VSWM systems was identified. The next step was to randomly choose a sample of thirty-five such neighborhoods from this population. We chose another thirty neighborhoods without the system. The survey instrument was pre-tested in a couple of neighborhoods in order to identify any shortcomings, difficulties in communication, and to estimate the time needed to go through the questionnaire, and then updated. Finally, the questionnaire was administered to the heads of households, and the initiators where applicable.

The two earliest VSWM systems in Dhaka were initiated by individuals who had been exposed to relatively more sanitary conditions than in Dhaka in the course of extended stays overseas. They were extremely motivated and put in large amounts of their own time and money in an effort to organize trash disposal in their neighborhoods and involve their neighbors in the effort. Many of the later systems that came into being were inspired by these pioneer systems, which had been publicized on television. The government can play a critical role in providing such publicity – and it may even be feasible for the government to put together a “how-to” guide

11

that would assist interested neighborhoods in profiting from the experience of the pioneers in terms of organizing their own SWM systems.

It is interesting to note that in the neighborhoods where VSWM systems exist, a fixed amount per household per month, irrespective of household size or any other consideration, is charged for the service. The monthly charge varies from $0.20 to $0.60 (with median $0.30), with higher charges when the collected garbage needs to be removed from the neighborhood altogether. The initial investment required to start such a system appears to be low, varying between $50.00 and $600.00, with a median of $280.00. While nearly all the initiators were unanimous in that the main problem in operating the systems was getting people to pay for a service that it is the city’s responsibility to provide, some VSWM systems have been able to run on a commercial basis (i.e., make a profit). This leads us to posit that the privatization of trash collection services in Dhaka city is feasible. The minimum scale required is around 250 households according to the initiators we interviewed. Given that SWM is the city’s responsibility one can argue for the provision of nominal tax breaks to neighborhoods which decide to organize and pay for their own trash collection. However the relative dearth of extant VSWM systems means a more detailed study would be required before we can make such a policy recommendation.

12

6. DATA DESCRIPTION Our dataset covers sixty-five neighborhoods and 652 households, with ten households on average from each neighborhood. The basic unit of analysis is the neighborhood. Neighborhood level variables such as median education, median per capita income, median tenure in the neighborhood, residents’ profession, etc., were calculated on the basis of household information collected during the survey. Information on the number of households in the neighborhoods, proportion of landlords and tenants among the residents, district of origin of residents, etc. was collected during the non-household interviews with initiators or knowledgeable persons from the community. Information on the divisional origin of residents was used as a dimension of community homogeneity. 15 Another aspect was provided by the proportion of resident landlords, as well as the percentage of adults in the neighborhood working in business, service, or professional work, etc. We presumed that older neighborhoods might be more likely to have stronger community ties and thus greater cooperative activity, and used neighborhood age, based on when the first residents moved in, as well as the average length of tenure to measure this.

Table 1 presents summary statistics of the variables used in our analysis and Table A1 in Appendix A presents the correlations between the variables. Table 1A provides an illustrative snapshot of the difference in mean levels of these variables in neighborhoods with and without VSWM systems.

15 The degree of homogeneity proxies the commonality of values and priorities in the neighborhood.

13

Table 1. Summary Statistics and Variable Definitions Variable Label N Mean Std. Dev. Minimum Maximum NHTRUST Avg trust score in observed HHs 65 1.49 0.32 1.00 2.43 NHRECIP Avg reciprocity score in observed

HHs 65 2.46 0.23 2.00 2.90

NHSHARE Avg sharing score in observed HHs 65 2.39 0.27 1.72 2.94 NHAGE 1998 – year when first resident

moved into NH 65 55.78 27.60 18.00 98.00

MDTENURE Median number of years respondents lived in NH

65 20.28 13.51 1.50 67.00

CMAX 1 if Chittagong is district with highest share

65 0.57 0.50 0.00 1.00

REGCONC Share of residents from district with highest share.

59 51.62 14.84 30.00 95.00

SHLNDLRD Share of residents that are landlords

65 25.64 6.35 16.00 40.00

BIS Share of NH jobs in business 65 23.03 11.30 0.00 50.00 MPCINC Median monthly per capita income

of respondents (Taka) 65 3394.97 1283.13 1200.00 8267.86

NHH Number of HH in NH 65 605.63 352.63 63 1500 INFRAST # centers, clubs, fields, and

meeting places in NH 65 0.631 0.977 0 4.00

PRIVPRV2 # sport, women’s orgs in NH before VSWM

65 0.385 0.490 0 1.00

OPBLPRV2 # religious, welfare, NH watch, library orgs in NH before VSWM

65 0.754 0.771 0 3.00

MDNHEDUC NH median of mean yrs of education of adults in HH

65 11.59 1.72 6.50 14.45

The survey provided several measures of social capital. This allows us to test the applicability of the theories of Coleman and Putnam to public good provision as well as to identify the differential effects of various types of social capital on cooperation. Following Knack and Keefer (1997) and others, we use a measure of trust, plus two unique measures of norms of reciprocity and sharing (based on our questionnaire) as proxies for social capital. Our measures for trust, reciprocity, and sharing are based on the mean of the categorical scores of the individual households related to the following questions in the questionnaire.

Trust: 1) Would you hire someone based on your neighbors’ recommendations? 2) In an emergency would you leave your young children with your neighbors?

Reciprocity: 1) Do you or your neighbors help arrange funerals for someone who dies in the neighborhood? 2) Do you or your neighbors send food to the family after a death in the family of your neighbors? 3) Do you or your neighbors help each other in taking sick neighbors to doctors or hospitals?

Sharing: 1) Do you or your neighbors send each other cooked food or sweets during religious and social festivals or on any happy occasion? 2) Do you or your

14

neighbors share fruits or vegetables grown on your own premises or village home?

In addition, we develop measures of activity in civic and social associations, which Putnam argues can be an important source of social ties that build social capital. Information on associations and the existence of various community facilities was derived from the non-household interview part of the survey.

We used two variables to proxy the associational depth of the community, both based on the number of associations or organizations that had existed prior to the establishment of the VSWM system to ensure exogeneity with respect to the VSWM system. The first was the number of associations providing a “private” good or service, where we counted sports and women’s associations – whose services are typically available to association members only. The second was the number of organizations providing a “public” good or service, where we counted welfare associations, neighborhood watches, library associations, and religious associations.

These two variables are measured with some error since we measure the age of an organization by the date when the earliest household joined it – so we only know the age of organizations if one of our respondents is a member. If the organization exists but no member of our sample belongs to it, we cannot count it since we cannot posit an age for it. In addition, when there are multiple organizations of the same type (e.g., sports clubs) in a neighborhood the data do not allow us to distinguish between them and thus only the oldest is likely to be counted. This said, we think the errors are unlikely to lead to a serious undercount and we have proceeded to use these variables as measures of associational activity.

Finally we included a measure of the physical infrastructure related to interaction between neighborhood residents – the number of meeting places, playing fields, etc. – on the thesis that this might encourage more interaction and thus build social capital in the neighborhood.

15

Table 1A. Comparison of Means Across Neighborhoods with and without VSWM Systems in Place

Variable

EXIST=1 EXIST=0

NHTRUST 1.6 1.33 NHRECIP 2.55 2.37 NHSHARE 2.41 2.3 REGCONC 45 60 NHH 650 502 SHLNDLRD 27.78 22.02 MDNHEDUC 12 11.22 MDTENURE 15.5 24.25 MDPCINC 3166.67 3237.5 BIS 20 26.82 PRIVPRV2 0 1 OPBLPRV2 1 0.5

16

7. RESULTS (i) Social Capital Estimates of the social capital regressions are presented in Table 2. Determinants of the measures of trust, reciprocity and sharing as estimated by least squares are listed in the first three columns. Each proxy for social capital was regressed against the dummy variable for the presence of a VSWM system (EXIST), previous participation in private good and public good oriented civic associations, median tenure of residents in the neighborhood, predominance of residents from Chittagong, share of residents that are landlords, share of business among neighborhood occupations, and the number of community social facilities.

Both the share of adults working in business and the share of residents that own their home are positively and significantly associated with all three measures of social capital. The business community in Dhaka does tend to be fairly close knit which may contribute to business’ effect. But the strength of the business variable says more than that. It also implies that members of the business community foster trust and norms of reciprocity among other community members, including those not involved in business. This suggests that people involved in business recognize some benefit from encouraging these kinds of ties between community members. Not surprisingly, homeowners appear to have a stronger effect on community social ties than tenants who may be more temporary residents of the neighborhood and who may have less invested there. Neighborhoods with a majority of residents who originate from Chittagong district also have stronger norms of reciprocity and sharing as their reputation suggests, although trust among residents is not significantly greater in these neighborhoods. Surprisingly, the number of community social facilities and meeting places builds norms of reciprocity, but is also negatively associated with trust. This variable may act as a proxy for the level of development of the neighborhoods since it measures physical infrastructure. Less-developed neighborhoods are likely to have more informal types of social ties that do not rely on structured meeting places, but that rely on, and perhaps engender, higher levels of trust.

17

Table 2. Determinants of Social Capital

NHTRUST NHRECIP NHSHARE NHSHARE Variable OLS OLS OLS 2SLS

INTERCEPT 0.8427** (0.1966)

1.8705** (0.1398)

1.4387** (0.1523)

1.4586** (0.2147)

EXIST 0.1348 (0.0839)

0.0916 (0.0596)

0.1022 (0.0650)

0.0165 (0.0395)

MDTENURE 0.0041 (0.0029)

0.0040* (0.0021)

0.0024 (0.0023)

0.0015 (0.0043)

CMAX 0.1388 (0.0889)

0.1641** (0.0632)

0.2580** (0.0688)

0.2434** (0.0947)

SHLNDLRD 0.0104* (0.0061)

0.0074* (0.0043)

0.0167** (0.0047)

0.0177** (0.0071)

BIS 0.0076** (0.0037)

0.0070** (0.0027)

0.0111** (0.0029)

0.0113** (0.0045)

INFRAST -0.1318** (0.0378)

0.0702** (0.0269)

0.0138 (0.0293)

0.0131 (0.0412)

PRIVPRV2 0.0736 (0.0841)

0.0274 (0.0598)

0.0915 (0.0652)

0.0853 (0.0844)

OPBLPRV2 0.0286 (0.0492)

-0.0499 (0.0350)

-0.0376 (0.0381)

-0.0196 (0.0509)

Adj. R2 0.2626 0.2611 0.3819 N 65 65 65 59 T-stat for exogeneity test 0.6050 1.4345 1.8791

* Significant at 10% ** Significant at 5%

It is interesting to note that participation in civic associations is not associated here with increased trust or stronger norms of reciprocity and sharing. To the uninitiated this appears to be a rejection of Putnam’s hypothesis that the kind of social ties developed in such associations is an important contributor to social capital.16 However, it is likely that formal membership in groups, which has been demonstrated to be a significant determinant of social capital in Putnam’s studies of the US and Italy, is not as important as other, more casual, forms of associational activity in Dhaka. Hence social capital formation in Dhaka is not seen to be dependent on participation in civic associations.

We conducted Hausman tests for the exogeneity of the EXIST variable in each of the social capital regressions.17 We are unable to reject the exogeneity of EXIST in the NHTRUST and NHRECIP equations, but we can reject the null hypothesis for the NHSHARE equation at the 7% level. We remove the bias in the EXIST parameter estimate in the NHSHARE equation by re-estimating the equation using the 2SLS technique developed by Amemiya (1978) for

16 However, as with other results from these regressions, we must recognize that the lack of significance in parameter estimates may be due in part to the small sample size for the neighborhoods in Dhaka. 17 The corresponding t-statistics for this parameter are posted in the last row of Table 2.

18

discrete choice simultaneous equations systems.18 Note that the effect of EXIST on NHSHARE is substantially reduced once the simultaneity has been removed.

Finally, we note that the presence of a VSWM system is not an important determinant of social capital as measured by trust and norms of reciprocity and sharing. This may not be surprising given that the majority of the trash disposal schemes were developed quite recently, in most cases less than three years before the survey. If the positive effect of the creation of such a system on social ties is not experienced all at once, but rather develops over time as the number of interactions associated with VSWM systems increase and their benefits become clearer, then the lack of explanatory power of this variable should be expected.

(ii) Existence of a VSWM System

The results of the estimation of the 2SCML regression for the formation of a VSWM system in the neighborhood are presented in Table 3. We estimate this probit using each of the three contemporaneous measures of social capital as regressors in separate equations and then estimate the probit again including all three measures in the same regression. The reduced form regressions used to generate the Vi terms are presented in Appendix Table A3.

The last row of Table 3 presents the value of the Wald statistic (Rivers and Vuong (1988)) used to test for the exogeneity of social capital variables in the cooperation regression for each of the regressions estimated. In each case, exogeneity is rejected, implying that the use of the 2SCML technique to obtain consistent parameter estimates is justified. The estimation results for the cooperation regressions in Table 3 show that the largest and most significant effect from a single social capital variable on the probability that a VSWM system will be formed is due to the strength of norms of reciprocity.

We calculated the marginal effect of an increase in NHRECIP on the probability that a VSWM system will form. Fixing all the other independent variables at their sample means, we obtain a marginal effect of 2.69 (standard error of 0.834). This suggests that an increase in the NHRECIP index of 0.1 units will lead to an increase in the probability that a VSWM system will be formed of 27 percent. Social capital indeed appears to have a major effect on cooperation for public good provision. The fact that of the three measures of social capital reciprocity has the greatest impact is consistent with the idea that reciprocity best represents the relationship underlying the phenomenon of organizing for SWM in the neighborhood.

18 The standard errors presented for the 2SLS regression in Table 2 are those resulting from the variance-covariance matrix derived by Amemiya, with the adjustment that the covariance of residuals from reduced form regressions for social capital and cooperation was calculated using the method described by Heckman (1978).

19

Table 3. Two-Stage Conditional Maximum Likelihood Estimates for EXIST Regression

Social Cap Vars. NHTRUST NHRECIP NHSHARE ALL 3 2SCML 2SCML 2SCML 2SCML NHTRUST 1.5045

(1.216) 0.1918

(2.737) NHRECIP 6.7371**

(2.092) 8.3312

(7.636) NHSHARE 5.6696**

(1.854) 0.5002 (6.567)

INTER -2.7586 (2.637)

-19.4028** (6.466)

-12.56** (4.875)

-24.7185** (12.01)

REGCONC -0.0489** (0.0193)

-0.0622** (0.0247)

-0.089** (0.0285)

-0.0828* (0.0429)

MDPCINC -0.0001 (0.0002)

-0.0002 (0.0003)

0.0000 (0.0002)

-0.0002 (0.0004)

MDNHEDUC 0.2902* (0.1727)

0.6199** (0.247)

0.3542 (0.217)

0.7397* (0.4086)

NHH -0.0001 (0.0006)

-0.0007 (0.0007)

0.0001 (0.0007)

-0.0009 (0.0014)

PRIVPRV2 -0.818* (0.4448)

-1.691** (0.6286)

-1.9872** (0.7968)

-2.5353** (1.007)

OPBLPRV2 0.5082 (0.3064)

0.9086** (0.3708)

0.6837 (0.4209)

1.2124* (0.6813)

V1H -1.236 (1.458)

-0.9142 (2.907)

V2H -4.4463* (2.344)

-6.1205 (7.998)

V3H -2.2779 (1.853)

3.5775 (6.719)

N 59 59 59 59 Modified Wald stat. for exogeneity of social capital

7.6762 65.316 144.9 2.08e+005

* Significant at 10% ** Significant at 5%

Norms of sharing also have a significant positive effect on the probability of a VSWM system being formed. The effect of the sharing variable is nearly as large as that of reciprocity with a marginal effect of 2.25 (standard error of 0.736). Trust on the other hand is not an important determinant of VSWM system formation. The marginal effect of trust is only 0.600, but this statistic is not significant, with a standard error of 0.485. We conjecture that the relatively low stakes involved, and the transactional nature of coordinated action for solid waste disposal may mean that trust between neighbors is not particularly important for setting up such systems. After all, the decision to participate does not imply a long-term binding commitment. Commonality of interests, as captured by reciprocity, may well be all that is required.

When the social capital variables are included together as regressors, the strength of the effect of social capital on the probability that a VSWM system will exist disappears. This may be due to multicollinearity among these measures. Indeed, the correlation between NHRECIP

20

and NHSHARE is 0.62, that between NHRECIP and NHTRUST is 0.37 and that for NHTRUST and NHSHARE is 0.44.

The median level of average household education in the neighborhood is also strongly related to the likelihood of a VSWM system coming into being in all equations except when social capital is measured by NHSHARE alone. There is a strong negative relationship between regional concentration in the origin of households in the neighborhood and the existence of a VSWM system. We conjecture that this may reflect closed-ness to new ways of doing things or new initiatives in a group that is very homogeneous. Median per capita income is not significantly associated with the existence of VSWM systems although the sign on income is negative in three of the four regressions. Although we were unable to control for the quality of municipal service provision in the neighborhoods prior to the establishment of VSWM systems, we suspect that the negative sign on income captures the quality of service delivery to an extent, making it less likely that higher income areas would need to organize for collective action. There are also no significant effects of neighborhood size on the probability that a VSWM system will form, suggesting that there are not strong coordination advantages available to smaller neighborhoods; nor are there benefits to having a larger pool of potential participants to choose from. Or perhaps these two effects are offsetting, making it difficult to identify the relationship between neighborhood size and the prospects for coordinated action for public good provision.

Most intriguing was the strong negative effect of the number of private good oriented organizations on the existence of VSWM systems, while the effect of the number of public good oriented associations is relatively insignificant, apart from in the NHRECIP equation. This may imply a sort of displacement or crowding out effect whereby the orientation of the “private” group militates against more publicly oriented activity. It is worth repeating that these variables count the number of groups or associations that existed PRIOR to the formation of a VSWM system, and are thus truly exogenous. Clearly it isn’t only that associational activity matters, but also that the type of associational activity is important.

21

8. CONCLUSIONS This paper has presented the results of a micro-empirical survey based study of households in 65 neighborhoods of Dhaka. Our results indicate that the organization of VSWM systems is a function of our proxy for social capital and of measures of associational activity, as well as the nature of such activity. Our results show that the different proxies we have used for social capital – trust, reciprocity and sharing – do, indeed, capture different aspects of social capital, with quite different impacts on community outcomes. Reciprocity among neighbors is far more important when it comes to cooperating for solid waste management than trust, for instance.

A clear policy implication of the analysis is that investments in education are likely to have spillover effects in terms of the ability to organize for SWM. However, we cannot say whether or not the promotion of associational activity (as has been mooted by proponents of the role of social capital in development following Putnam) has a positive net impact on the provision of a public good or service by a neighborhood committee – either directly or through higher social capital. We need also to remember the insight of Beall’s (1997) case studies, that the nature of the good or service, and the circumstances of its underprovision should not be forgotten.

The different aspects of social capital do not appear to be determined by the existence of a VSWM system, though they are in part explained by some of the other structural and associational variables in our dataset. Homogeneity of interests and points of view appear paramount in explaining levels of social capital. Thus from a policy viewpoint, social capital may well be a primal variable that resides in the inherent ability of different individuals to relate to one another. While it can be channeled to different uses, our analysis does not indicate that it is something policy makers can easily affect.

In terms of policy implications, this work shows the feasibility of privatizing the collection of solid waste in urban middle-class neighborhoods. It also indicates that the government can play a useful role in publicizing success stories and in providing information on how successful neighborhoods were able to organize themselves for SWM. However, more analysis is needed before one can conclude that the city is ready for wholesale privatization of the SW collection service. The most important policy implication of our work is that the introduction of public-private partnerships or self-help schemes is more likely to be successful in neighborhoods which are high in social capital. Thus social capital proxies or determinants can be used as predictors of success when targeting neighborhoods for different social or public good oriented interventions.

22

REFERENCES Amemiya, Takeshi. The Estimation of a Simultaneous Equation Generalized Probit Model.

Econometrica (1978) 46(5): 1193-1205. Axelrod, Robert. Complexity of Cooperation: Agent-Based Models of Competition and

Collaberation. Princeton: Princeton University Press, 1997.

———. Evolution of Cooperation. New York: Basic Books, 1984.

Baland, Jean-Marie, and Jean-Philippe Platteau. "Coordination Problems in Local-Level Resource Management." Journal of Development Economics 53 (1997): 197-210.

———. "Division of the Commons: A Partial Assessment of the New Institutional Economics of Land Rights." American Journal of Agricultural Economics 80 (1998): 644-50.

———. Halting Degradation of Natural Resources: Is There a Role for Rural Communities? New York: Oxford University Press, 1996.

Banerjee, Abhijit V. "A Theory of Misgovernance." Quarterly Journal of Economics 112, No. 4 (November 1997): 1289-332.

Beall, Jo. "Social Capital in Waste--A Solid Investment?" Journal of International Development 9, No. 7 (1997): 951-61.

Besley, Timothy, and Anne Case. "Modelling Technology Adoption in Developing Countries." American Economic Review 83, No. 2 (1993): 396-402.

Black, Jane, Maurice D. Levi, and David de Meza. “Creating a Good Atmosphere: Minimum participation for Tackling the ‘Greenhouse Effect’.” Economica 60, No. 239 (1993): 281-93.

Borjas, George J. "Ethnic Capital and Intergenerational Mobility." Quarterly Journal of Economics 107, No. 1 (1992): 123-50.

———. "Ethnicity, Neighborhoods, and Human Capital Externalities." American Economic Review 85, No. 3 (1995): 365-90.

Boxman, Ed A. W., Paul M. De Graaf, and Hendrik D. Flap. "The Impact of Social and Human Capital on the Income Attainment of Dutch Managers." Social Networks 13 (1991): 51-73.

Brown, L. David, and Darcy Ashman. "Participation, Social Capital, and Intersectoral Problem Solving: African and Asian Cases." World Development 24, No. 9 (1996): 1467-79.

Coleman, James S. Foundations of Social Theory. Cambridge, MA: Harvard University Press, 1990.

23

———. "Social Capital in the Creation of Human Capital." American Journal of Sociology 94 (1988): S95-S120.

Collier, Paul. "Social Capital and Poverty." Mimeo: Centre for Study of African Economies, University of Oxford, March, 1998.

de Janvry, Alain, Nancy McCarthy, and Elisabeth Sadoulet. "Endogenous Provision and Appropriation in the Commons." American Journal of Agricultural Economics 80 (1998): 658-64.

Evans, Peter. "Government Action, Social Capital and Development: Reviewing the Evidence on Synergy." World Development 24, No. 6 (1996): 1119-32.

Fox, Jonathan. "How Does Civil Society Thicken? The Political Construction of Social Capital in Rural Mexico." World Development 24, No. 6 (1996): 1089-103.

Fukuyama, Francis. Trust: The Social Virtues and the Creation of Prosperity. New York: The Free Press, 1995.

Granovetter, Mark. "The Strength of Weak Ties." American Journal of Sociology 78 (1973): 1360-1380.

Greif, A. "Cultural Beliefs and the Organization of Society: A Historical and Theoretical Reflection on Collectivist and Individualist Societies." Journal of Political Economy 102, No. 5 (1994): 912-50.

Harrison, Lawrence E. Who Prospers? How Cultural Values Shape Economic and Political Success. New York: Basic Books, 1992.

Harriss, John, and Paolo De Renzio. "'Missing Link' or Analytically Missing?: The Concept of Social Capital, An Introductory Bibliographic Essay." Journal of International Development 9, No. 7 (1997): 919-37.

Heckman, James J.. Dummy Endogenous Variables in a Simultaneous Equations System. Econometrica (1978) 46(4): 931-959.

Helliwell, John F., and Robert D. Putnam. "Economic Growth and Social Capital." Eastern

Economic Journal 21, No. 3 (Summer 1995): 295-307.

Isham, Jonathan, Kaufman D., and Lant Pritchett. "Civil Liberties, Democracy, and the Performance of Government Projects." World Bank Economic Review 11, No. 2 (1995): 219.

Jacobs, Jane. The Death and Life of Great American Cities. New York: Vintage Books, 1961.

Knack, Stephen, and Philip Keefer. "Does Social Capital Have an Economic Payoff? A Cross-Country Investigation." Quarterly Journal of Economics 112, No. 4 (November 1997): 1251-88.

24

Lin, Nan, and Mary Dumin. "Access to Occupations Through Social Ties." Social Networks 8 (1986): 365-85.

Lin, Nan, John C. Vaughn, and Walter M. Ensel. "Social Resources and Occupational Status Attainment." Social Forces 59 (1981): 1163-81.

Meyerson, Eva M. "Human Capital, Social Capital and Compensation: The Relative Contribution of Social Contacts to Managers' Income." Acta Sociologica 37 (1994): 383-99.

Moser, Caroline. "Confronting Crisis: A Comparative Study of Household Responses to Poverty and Vulnerability in Four Poor Urban Communities." Environmentally Sustainable Development Studies and Monographs Series No. 8. Washington, DC: The World Bank, 1996.

Narayan, Deepa, and Lant Pritchett. "Cents and Sociability: Household Income and Social Capital in Rural Tanzania." World Bank Policy Research Working Paper, July 1997.

Olson, Mancur. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, Mass.: Harvard University Press, 1965.

———. The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities. New Haven: Yale University Press, 1982.

Ostrom, Elinor. "Crossing the Great Divide: Coproduction, Synergy, and Development." World Development 24, No. 6 (1996): 1073-87.

———. Governing the Commons: The Evolution of Institutions for Collective Action. New York: Cambridge University Press, 1990.

Posner, R. A. "A Theory of Primitive Society, With Special Reference to Law." Journal of Law and Economics 23 (1980): 1-53.

Putnam, Robert D. "Bowling Alone: America's Declining Social Capital." Journal of Democracy 6, No. 1 (1995): 65-78.

———. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press, 1993a.

Putzel, James. "Accounting for the 'Dark Side' of Social Capital: Reading Robert Putnam on Democracy." Journal of International Development 9, No. 7 (1997): 939-49.

Rivers, Douglas, and Quang H. Vuong. “Limited Information Estimators and Exogeneity Tests for Simultaneous Probit Models”. Journal of Econometrics 39 (1988): 347-366.

Runge, C. Ford. "Common Property Externalities: Isolation, Assurance, and Resource Depletion in a Traditional Grazing Context." American Journal of Agricultural Economics 63 (November 1981): 595-606.

25

———. "Stategic Interdependence in Models of Property Rights." American Journal of Agricultural Economics 66 (December 1984): 807-13.

Sandler, Todd. Collective Action: Theory and Applications. Ann Arbor: University of Michigan Press, 1992.

Seabright, Paul. "Managing Local Commons: Theoretical Issues in Incentive Design." Journal of Economic Perspectives 7, No. 4 (1993): 113-34.

Sugden, Robert. Economics of Rights, Cooperation and Welfare. New York: Blackwell, 1986.

———. "Reciprocity: The Supply of Public Goods Through Voluntary Contributions." Economic Journal 94, No. 376 (December 1984): 772-87.

Tiffen, Mary. More People, Less Erosion: Environmental Recovery in Kenya. New York: J. Wiley, 1994.

Tocqueville, Alexis de. Democracy in America . New York: G. Adlard, 1838.

White, T. Anderson, and C. Ford Runge. "Common Property and Collective Action: Lessons From Cooperative Watershed Management in Haiti." Economic Development and Cultural Change 43, No. 1 (October 1994): 1-41.

27

APPENDIX A Table A1. Matrix of Pearson Correlation Coefficients

Variable NHTRUST NHRECIP NHSHARE NHAGE MDTENURE REGCONC SHLNDLRD BIS MDPCINC NHH INFRAST PRIVPRV2 OPBLPRV2 MDNHEDUC

NHTRUST 1.00 NHRECIP 0.37 1.00 NHSHARE 0.44 0.62 1.00 NHAGE -0.11 0.06 -0.05 1.00 MDTENURE 0.12 0.22 0.12 0.58 1.00 REGCONC 0.02 0.06 0.15 0.52 0.53 1.00 SHLNDLRD 0.25 0.22 0.39 -0.10 0.04 0.29 1.00 BIS 0.13 0.17 0.18 0.38 0.30 0.32 -0.18 1.00 MDPCINC -0.15 -0.03 -0.13 0.17 0.10 0.01 -0.31 0.06 1.00 NHH 0.01 0.08 -0.04 0.00 0.02 -0.09 -0.28 0.09 0.34 1.00 INFRAST -0.35 0.31 0.06 0.14 0.15 0.00 -0.03 0.10 -0.06 0.13 1.00 PRIVPRV2 0.00 0.11 0.09 0.23 0.32 0.17 -0.27 0.27 0.16 0.22 0.27 1.00 OPBLPRV2 0.21 0.06 0.16 -0.12 0.17 -0.01 0.15 0.20 -0.15 0.23 0.08 0.21 1.00 MDNHEDUC -0.01 -0.09 -0.01 -0.22 -0.39 -0.35 -0.10 -

0.38 0.46 0.29 -0.16 0.03 -0.24 1.00

28

Table A2. First-Stage Probit to Predict EXIST Dep. Var: EXIST (N=59)

Estimate Std. Error Wald Chi-Sq.

Prob

INTERCPT -5.704 3.8344 2.2129 0.1369 MDTENURE -0.0315 0.0332 0.8955 0.344 CMAX 0.9697 0.6545 2.1956 0.1384 SHLNDLRD 0.211 0.0795 7.0365 0.008 BIS 0.077 0.0367 4.4082 0.0358 INFRAST 0.4221 0.3166 1.7774 0.1825 REGCONC -0.1119 0.0452 6.1366 0.0132 MDPCINC -0.00011 0.00027 0.1671 0.6827 MDNHEDUC 0.3674 0.2768 1.762 0.1844 NHH 0.000497 0.000788 0.3974 0.5284 PRIVPRV2 -1.0125 0.6944 2.1262 0.1448 OPBLPRV2 0.5401 0.4044 1.7835 0.1817

29

Table A3. First-Stage Reduced Form Regressions

Dep Var: NHTRUST Dep Var: NHRECIP Dep Var: NHSHAR

E

R2: 0.40993 R2: 0.36562 R2: 0.47116 adjR2: 0.27183 adjR2: 0.21715 adjR2: 0.34738 DW: 1.9641 DW: 2.0876 DW: 2.4456

Variable Estim StdErr tstat Prob Estim StdErr tstat Prob Estim StdErr tstat Prob INTER 0.8027 0.4412 1.8193 0.07524 1.6953 0.3319 5.1083 5.83E-06 0.8329 0.3571 2.3324 0.02402 MDTENURE 0.0061 0.003449 1.7594 0.08501 0.0039 0.002595 1.5039 0.1393 0.0027 0.002792 0.9821 0.3311 CMAX 0.0672 0.09457 0.7101 0.4812 0.1722 0.07114 2.4205 0.01942 0.2954 0.07654 3.8597 0.000345 SHLNDLRD 0.0169 0.007547 2.2438 0.02959 0.0129 0.005677 2.279 0.02725 0.0151 0.006109 2.4796 0.0168 BIS 0.0082 0.004055 2.0103 0.05015 0.0077 0.00305 2.5394 0.01447 0.0124 0.003282 3.7865 0.000433 INFRAST -0.1401 0.03984 -3.5175 0.000978 0.0695 0.02997 2.3204 0.02471 0.0348 0.03225 1.0803 0.2855 REGCONC -0.0042 0.003542 -1.1981 0.2369 -0.0012 0.002664 -0.4322 0.6676 0.0032 0.002867 1.1076 0.2737 MDPCINC -0.0001 3.70E-05 -1.8834 0.06584 0 2.78E-05 -0.5053 0.6157 0 3.00E-05 -1.2592 0.2142 MDNHEDUC 0.0275 0.03295 0.8342 0.4084 0.009 0.02479 0.3633 0.718 0.0504 0.02667 1.89 0.06493 NHH 0.0001 0.00012 0.9837 0.3303 0.0001 9.03E-05 1.3767 0.1751 0 9.71E-05 0.2229 0.8246 PRIVPRV2 0.0258 0.09096 0.2836 0.7779 0.0085 0.06842 0.1248 0.9012 0.0083 0.07362 0.1133 0.9103 OPBLPRV2 0.0311 0.05578 0.5577 0.5797 -0.0584 0.04196 -1.3928 0.1702 -0.0068 0.04515 -0.1507 0.8809

30

APPENDIX B. NOTES ON ESTIMATION (B.1) Rivers and Vuong use the result that conditional maximum likelihood estimation is often the

most feasible estimation procedure when the joint density function of endogenous variables factors

into a marginal distribution and a conditional distribution. In this case, the convenient factorization

results in a marginal likelihood expression for the social capital regressions and a conditional

likelihood for the binary VSWM system variable.

(B.2) How does this method of estimation of equation (4.b) generate consistent parameter

estimates? First, after adding the term iV (estimated by iV̂ ) as in (5), if 0=+ λγ and 0≠λ (5)

becomes

iiii XXya ηλβ +Π′+′=∗ ˆ ).5( 1 .

This is identical to the instrumental variables approach to estimation for a discrete choice

simultaneous equations system of Heckman (1978) in which iY is replaced by its predicted value.

However, dropping these restrictions, we can rewrite (5) as

iiii XYyb ηνµ +′+′=∗ ).5( ,

where λγµ += and λβν Π−= ˆJ . In matrix notation,

ηνµ ++=∗ XYy ,

where ∗y is an 1×n vector, Y and X are mn × and pn × matrices, respectively, and the parameters

µ and ν are just as before. Premultiplying by ( ) XXX ′′ −1 ,

( ) ( ) ηνµ ++′′=′′ −∗− YXXXyXXXc 11 ).5( .

GLS estimation of (5.c) can be used to recover consistent estimates of µ and ν. Let ω+Π′=∗ Xy ~

be the reduced form cooperation regression. Then, substituting for reduced form parameter estimates

and ν, we have

ηλβµ +Π−+Π=Π ˆˆ~̂ J ,

which can be simplified as

(5.d) ηβγ ++Π=Π Jˆ~̂.

31

This is identical to the equation used by Amemiya to recover consistent estimates of the structural

parameters from reduced form parameter estimates for the discrete choice simultaneous equations

system. Amemiya’s approach involved GLS estimation of (5.d) where ( )Π−Π−

Π−Π= ˆ~~̂ γη .

So estimation of (5) above indeed yields unbiased estimates ( )βγ ˆ,ˆ .

(B.3) Since Rivers and Vuong are only concerned about efficient estimation of the discrete choice

equation in the simultaneous system in (4) above, they estimate regressions for the continuous

endogenous variables in reduced form, ignoring the potential effects of the latent variable

(cooperation) on these other endogenous variables (social capital).

(B.4) As noted above Heckman (1978) and Amemiya (1978) offer alternative approaches to

estimating discrete choice simultaneous equation models. Amemiya showed that his two-stage

estimator in (5.d), in which the structural parameters are recovered by regressing reduced form

parameter estimates from one equation against those from another, was more efficient than the

Heckman IV-type estimator in (5.a). Rivers and Vuong (1988) show that their estimator is more

efficient than the Amemiya estimator under certain conditions on the parameters, but not always.

(B.5) One of the benefits of the 2SCML method for estimating the cooperation equation (4.b) is that

it provides a simple method for testing exogeneity of the social capital variables in this equation.

Substituting for Vi in equation (5) and gathering terms,

iiiii XXYy ηλβµ +Π′+′+′= ˆ1

* ,

where λγµ += . This is the form of regression typically used to perform a Hausman test for

exogeneity of regressors in which the potentially endogenous variables are included along with their

predicted values. A test of the null hypothesis of exogeneity is equivalent to testing whether 0=λ .

Rivers and Vuong (1988) show that the modified Wald statistic given by

( ) λλλ ˆˆˆˆ 1−′= VnMW

has an asymptotic central chi-square distribution with m degrees of freedom.

social capital in solid waste management: evidence from dhaka

Documents