FREE ELECTRONIC LIBRARY - Dissertations, online materials

Pages:   || 2 | 3 | 4 | 5 |   ...   | 6 |


-- [ Page 1 ] --




Abstract. We examine how participation in a microfinance program diffuses through

social networks, using detailed demographic, social network, and participation data from

43 villages in South India. We exploit exogenous variation in the importance (in a net- work sense) of the people who were first informed about the program, the “injection points.” Microfinance participation is significantly higher when the injection points have higher eigenvector centrality. We also estimate structural models of diffusion that al- low us to (i) determine the relative roles of basic information transmission versus other forms of peer influence, and (ii) distinguish information passing by participants and non- participants. We find that participants are significantly more likely to pass information on to friends and acquaintances than informed non-participants. However, information passing by non-participants is still substantial and significant, accounting for roughly one-third of informedness and participation. We also find that, once we have properly conditioned on an individual being informed, her decision to participate is not signifi- cantly affected by the participation of her acquaintances.

JEL Classification Codes: D85, D13, G21, L14, O12, O16, Z13 Keywords: Microfinance, Diffusion, Social Networks, Peer Effects Date: April 2012 (NBER Working Paper 17743).

We gratefully acknowledge financial support from the NSF under grants SES-0647867, SES-0752735 and SES-0961481. Chandrasekhar thanks the NSF GRFP for financial support. Daron Acemoglu and various seminars provided helpful comments and suggestions. We also thank Ben Feigenberg, Randall Lewis, Bryan Plummer, Gowri Nagaraj, Jeff Guo, Tomas Rodriguez-Barraquer, Xu Tan, and Adam Sacarny. The Centre for Microfinance at the Institute for Financial Management and Research and BSS provided valuable assistance.

† Massachusetts Institute of Technology.

Stanford University, the Santa Fe Institute, and CIFAR.


1. Introduction Information is constantly being passed on through social networks: friends get pure information from friends (for example, they might learn about the existence of a new product) as well as opinions (for example, on whether or not the the product works).

While there are numerous studies documenting such phenomena,1 few model the exact mechanics of information transmission and empirically distinguish between alternative models of transmission. However, understanding how information exchangetakes place is crucial to the design of effective campaigns, for example in public health. In this paper, we use rich data that we collected and a combination of structural and reduced-form approaches to investigate the nature of information exchange.

The data include detailed information about social networks from 75 different rural villages in southern India as well as information on the subsequent diffusion of microfinance participation in 43 of those villages. The data are unique given the large number of different villages for which we have observations, the wealth of information on possible connections (we have data on 13 different types of relationships, from whether respondents go to the temple together to whether they borrow money or kerosene from one another), and the fact that the data are matched with administrative data on the take up of microfinance in 43 of these villages, collected over a period of more than a year.

Our analysis consists of two main components that differ in both the issues explored and techniques employed. The first is a reduced-form approach in which we take advantage of cross-village variation in network characteristics and initial contact nodes to identify what influences diffusion. The second is a structural modeling approach in which we explicitly model information passing from household to household within networks, and subsequent participation decisions. This allows us to infer the importance of pure information transmission about microfinance availability relative to opinion transmission and peer effects. Next, we describe each of the two approaches in more detail.

In our reduced-form analysis, we test which attributes of network structure are significantly related to microfinance diffusion. One of the main concepts that we analyze is the role of the initial injection points. Specifically, if only ten or twenty members of a village of a thousand people are initially informed about the opportunity to borrow from a micronance institution, how does eventual participation depend on exactly which individuals 1 The literature documenting diffusion in various case studies spans decades from Ryan and Gross (1943) on the diffusion of hybrid corn adoption, to Lazarsfeld et al. (1944) on word-of-mouth influences on voting behavior, to Katz and Lazarsfeld (1955) on the roles of opinion leaders in product choices, to Coleman et al. (1966) on connectedness of doctors and new product adoption. More recently the literature includes Foster and Rosenzweig (1995) and Conley and Udry (2010) on learning and agricultural technology, and Kremer and Miguel (2007) on deworming drugs; this literature includes both empirical and theoretical analyses. For background discussion and references, see Rogers (2003), Jackson (2008), and Jackson and Yariv (2010).


are initially contacted? The previous empirical literature is largely silent on this topic, although some case studies and theory speak to this.2 The setting we examine is particularly suitable for studying this question because our microfinance partner always follows the same procedure for informing a village about microfinance opportunities: the partner identifies specific people in the village (teachers, shopkeepers, etc.) and calls them the “leaders” (irrespective of whether they are, in fact, opinion leaders in that particular village), informs them about the program, and asks them to tell other potentially interested villagers about the program. This fixed rule provides exogenous variation across villages in the network characteristics of the individuals who were initially contacted, which we show are uncorrelated with other attributes of the village. In some villages, those who were initially contacted are more centrally positioned in the network than in other villages. We show that eventual program participation is higher in villages where the first set of people to be informed are more important in a network sense. In particular, we show that a specific measure of the importance or position of the initially contacted individuals, their eigenvector centrality (explained below), is significantly related to eventual microfinance participation, while other measures of centrality are not.

We also examine the effects of other village level measures of network connectivity, such as average degree, average path length, clustering, etc., which capture the characteristics of the network as a whole, rather than the network position of the injection points. While there are theoretical arguments suggesting that a number of these characteristics could play substantial roles in determining the nature of diffusion, we do not find significant evidence that they do in this setting (and explain why this might be).

The second major contribution of the paper is to develop and structurally estimate models that distinguish alternative mechanisms for the diffusion of information. Here, we move beyond the existing literature in two ways.

First, the models that we introduce allow for information to be transmitted even by those who are informed but choose not to participate themselves (and we allow nonparticipants to transmit information at a different rate from participants). This contrasts with standard contagion-style diffusion models, in which the diffusion is modeled as an infection: an individual needs to have infected neighbors to become infected. In our model, people who become informed and are either ineligible or choose not to participate can still tell their friends and acquaintances about the availability of microfinance; and, in fact, we find that the role of such non-participants is substantial and significant. We also find that there is a significant participation effect in information transmission: people who do participate are estimated to be more than four times as likely to pass on information about microfinance to their friends as non-participants. Nonetheless, non-participants transmit a significant amount of information, especially since there are many more non-participants 2 See Jackson and Yariv (2010) as well as the discussion in Section 3.2 for references and background.


in the village than participants. In fact, our estimates indicate that information passing by non-participants is responsible for one-third of overall information about the program and program participation.

Second, in our framework, whether a person participates in microfinance can depend on whether the person is aware of the opportunity (a pure information effect), as well as whether the person’s friends and acquaintances themselves participate in the program (which we loosely deem an “endorsement effect”). Note that endorsement here is a catchall for any interaction beyond basic information transmission: such effects might be driven by complementarities, by substitution, by imitation, by opinion transmission, etc. Diffusion models generally focus on one or the other of these effects, and we know of no previous study that empirically distinguishes between them. Indeed, without a structural model, these effects are difficult to distinguish since they have similar reduced-form implications.

In both cases, friends of people who take up microfinance will be more likely to take it up themselves than friends of those who do not take up microfinance.

By explicitly modeling the communication and decision processes as a function of network structure and individual characteristics, we are able to separately estimate information and endorsement effects. We find that the information effect is significant. Once informed, however, an individual’s decision is not significantly influenced by the fraction of her friends who participate. In this sense, we find no (statistical) evidence of an endorsement effect (once one allows for differential information passing rates for takers and non-takers).3 Of course, the usual challenges of identifying diffusion models remain: the social networks are endogenous and there tend to be strong similarities across linked individuals, which could generate correlations in their decisions even when there is no diffusion. To explore the empirical importance of such correlations, we compare our model of information transmission with an extended version of the model in which there is an added component of correlation. In the extension, take up is also allowed to be a function of an individual’s network distance from initially informed households who choose to participate, (say) because of similarities between individuals who are at similar distances from the leaders. We show that this extended model of information transmission that that incorporates distance from leaders does not substantively change our estimates of the information transmission parameters by participants and non-participants. Moreover, we estimate our main model on an alternative set of moments exploiting a different set of variation and show that our conclusions are not substantively altered. As a final check, we show that the model does well in predicting aggregate patterns of diffusion over time, 3 Note that this is different from distinguishing peer effects from homophily, whereby peer effects are diminished when one properly accounts for the characteristics of an individual and the correlation of those characteristics with those of his or her peers (e.g., see Aral et al., 2009). Here, the endorsement effect disappears when we separate out information transmission from other diffusion effects.


even though the data used for the estimation of the model are only the initial injections points and the final take-up patterns.

The remainder of the paper is organized as follows. In Section 2 we provide background information about our data. Section 3 outlines our conceptual framework. Section 4 contains the reduced-form analysis of how network properties and initial injection points correlate with microfinance participation. In Section 5 we structurally estimate diffusion models that distinguish the impacts of information transmission, endorsement effects, and simple distance from injection points on patterns of microfinance participation. Section 6 concludes.

2. Background and Data

2.1. Background. This paper studies the diffusion of participation in a microfinance program run by Bharatha Swamukti Samsthe (BSS) in rural southern Karnataka.4 BSS operates a conventional group-based microcredit program: borrowers (women only) are formed into groups of 5 and are jointly liable for their loans. The starting loan is approximately 10,000 rupees (just over 200 dollars) and is repaid in 50 weekly installments.

The annualized interest rate is approximately 28%.

When BSS starts working in a village, it seeks out a number of pre-defined leaders, who BSS expects to be well-connected within the village: teachers, leaders of self-help groups, and shopkeepers. BSS first holds a private meeting with the leaders. At this meeting, credit officers explain the program to the village leaders, and then ask them to help organize a meeting to present information about microfinance to the village, and to tell their friends about microfinance. These leaders play an important part in our identification strategy, since they function as injection points for microfinance in the village. After BSS meets with village leaders, interested, eligible villagers (women between the ages of 18 and 57) contact BSS, are trained and formed into groups, and credit disbursements begin.

Pages:   || 2 | 3 | 4 | 5 |   ...   | 6 |

Similar works:

«evaluation of coordination and complementarity of european assistance to local development annexes Commissioned by: Swedish International Development Cooperation Agency Austrian Development Agency Belgian Federal Public Service Foreign Affairs, Foreign Trade and Development Cooperation – Directorate General for Development Cooperation French Ministry of Foreign Affairs Irish Aid The Development Cooperation of the Netherlands Evaluation of coordination and complementarity of European...»

«Theorising the links between HR and Governance: issues and challenges for HRD Graeme Martin Department of Management University of Glasgow Glasgow, G12 8QQ g.martin@mgt.gla.ac.uk 07777637059 Jim McGoldrick (corresponding author) NHS Fife Board Hayfield House Hayfield Road Kirkcaldy, KY2 5AH james.mcgoldrick@nhs.net and jim.mcgoldrick1@btopenworld.com 01592 648080 and 07748 731921 Stream: Critical Perspectives in Human Resource Development. Paper type: Refereed Track Key Words: Corporate...»

«AUTO THEFT IN CENTRAL PHILADELPHIA by George F. Rengert Temple University Abstract: Auto theft is a major concern for cities attempting to revitalize their centers. Tourists, shoppers and workers will avoid this region of the city if they think their property may be stolen. This research is an attempt to identify whether public and private facilities about which parked cars are expected to be clustered at specific times of the day provide a focus for auto thieves. Using the Spatial and Temporal...»

«1 Università degli Studi di Roma Tor Vergata Dipartimento di Ingegneria Civile e Ingegneria Informatica GeoInformation PhD Program Application of KLIMA algorithm to Carbon Dioxide retrieval from IASI/METOP-A observations and comparison with TANSO-FTS/GOSAT products Lucia Maria Laurenza A.A. 2013-2014 Phd Mentor: Prof. Giovanni Schiavon IFAC-CNR supervisor: Dr. Ugo Cortesi 2 Table of Contents 1 Introduction 5 1.1 Atmospheric Carbon Dioxide and Global Warming.....................»

«Color record in self-monitoring of blood glucose improves glycemic control by better self-management (カラー記録を活用した血糖自己測定は自 己管理行動と血糖コントロールの改善に寄 与する) 西村 亜希子 【主論文】 JOURNAL: Diabetes Technology and Therapeutics Volume 16, Number 7, 2014 © Mary Ann Liebert, Inc. DOI: 10.1089/dia.2013.0301 TITLE: Color record in self-monitoring of blood glucose improves glycemic control by better self-management...»

«PROJECT DATABASE FOR SUB-SAHARAN AFRICAN COUNTRIES Prepared by Africa Project Access November 2013 1 Guide to Using the Africa Project Database 1. Africa Project Access provides two essential back-up services to the Projects supplied to subscribers via the Africa Project Database. In order to streamline the usage of these services, subscribers are kindly requested to refer to the Date and “Description” of each Project concerned. (There is therefore no need for an elaborate numbering...»

«Connections Day 2016 | Question & Answer Academics, Classwork, and Grades | p. 1 Alcohol and Other Drugs | p. 3 Athletics | p. 4 Cars and Parking | p. 5 Clubs and Organizations | p. 5 Greek Life | p. 6 Orientation | p. 7 Recreation | p. 8 Residence Life and Housing | p. 9 Safety | p. 11 Work Study and Jobs After Redlands | p. 14 Additional Resources and Miscellaneous Questions | p. 14 Academics, Classwork, and Grades After you use your two included May Terms can you still go on a May Term study...»

«Christopher Middleton studied at Oxford and then taught at the University of Zürich, at King’s College, London, and finally as Professor of Germanic Languages at the University of Texas, Austin. He has published translations of Robert Walser, Nietzsche, Hölderlin, Goethe and many contemporaries, receiving several awards, including the Geoffrey Faber Memorial Prize and the Schegel-Tieck Translation Prize. His poems, essays and selected translations are all published in the UK by Carcanet...»

«ALBOLD, CHERYLL A., Ph.D. Emerging Scholars: Predictors of Independent Researcher Identification in Education PhD Students. (2011) Directed by Dr. Deborah J. Taub. 196 pp. Conducting independent research is a rite of passage in most PhD programs. Yet, little is known about the factors in the higher educational process that meaningfully shape doctoral students‘ self-beliefs about conducting independent research. This study used psychosocial and social cognitive variables to predict doctoral...»

«CURRICULUM VITAE UMAPADA PAL, Ph. D. Residencial Address: 57 Poniente 1304-2, Col. Prados Agua Azul, Puebla, Pue. 72430, Mexico. Tel.: +52-222-2439125. E-mail: umapadapal@hotmail.com Midnapore (West Bengal, India), 23rd January, 1960.Place and date of birth: Nationality: Indian Merital status: Married (with two children) CURP : PAXU600123HNELXM03 RFC: PAXU6001235P9 Language ability: English (100%), Spanish (90%), Bengali (100%), Hindi (70%) Profesor Investigador Titular ‘C’ (Full professor,...»

«In the Name of the Son (and the Daughter): Intergenerational Mobility in the United States, 1850-1940.∗ Claudia Olivetti and M. Daniele Paserman Boston University and NBER January 2015 Abstract This paper provides a new perspective on the evolution of intergenerational mobility in the United States in the late 19th and early 20th centuries. We devise an empirical strategy that allows us to estimate trends in intergenerational elasticities between fathers and children of both sexes. The key...»

«252 SEDER OLAM AND THE SABBATICALS ASSOCIATED WITH THE TWO DESTRUCTIONS OF JERUSALEM PART II RODGER C. YOUNG In the previous issue of this journal, the first of the two articles in this series established that the Seder Olam of Rabbi Yose ben Halaphta stated that the two destructions of Jerusalem were both in the latter part (motzae) of a Sabbatical year. Those Sabbatical years began, respectively, in Tishri of 588 BCE and Tishri of 69 CE. This understanding is compatible with the proper...»

<<  HOME   |    CONTACTS
2016 www.dissertation.xlibx.info - Dissertations, online materials

Materials of this site are available for review, all rights belong to their respective owners.
If you do not agree with the fact that your material is placed on this site, please, email us, we will within 1-2 business days delete him.