Doctoral Dissertation Research: Self-Help and Self-Government: How Local Governments Affect Microcredit Provision in South India

该项目调查了小额信贷群体（获得小额、无担保、连带责任贷款的低收入妇女群体）与印度南部农村地方人民政府之间的互动。来自有限数据集的初步结果表明，当地政治领导人的性别对其选区内小额信贷团体获得的信贷数量有很大影响；这令人费解，因为这些领导人在提供这些贷款方面没有正式的角色。本项目使用从印度泰米尔纳德邦的一个行政部门收集的数据来检验关于领导人增加贷款的机制的不同假设，以及为什么女性领导人比男性领导人增加贷款的动机。该项目的核心包括两个调查，一个是对所有小组领导的调查，另一个是对每个小组随机抽取的5名成员的调查。领导调查引出了该集团在过去10年中收到的贷款信息；集团的其他一般资料，例如成立年份及与集团有关联的银行；还有一份小组成员名单。该名册构成了成员调查的样本框架，从中获得有关小组成员家庭结构的信息；她的家庭通过政府计划获得的福利的资料；以及她与当地政治领导人关系的信息。这些调查还补充了每个选区每一届的选举数据，包括政府准随机为女性候选人保留的选区的数据。利用这些数据，我们有可能检验这样一个假设：女性领导人通过缓解银行对这些团体可能拖欠贷款的担忧，增加了她们所在选区的贷款，而且她们比男性领导人在这方面做得更多，因为由于印度的贷款补贴结构，银行向女性领导人支付报酬的成本更低。通过测试，看看补贴率的波动是否对男性和女性领导下的无补贴贷款有不同的影响，有可能排除一个竞争性的假设，即女性领导在她们的团队中获得更多的信贷，因为她们比男性领导对那些团队更无私。如果女性领导下的无补贴贷款比男性领导下的无补贴贷款对补贴增加的反应更积极，这就是反对利他主义假说的证据。这个项目具有重大的智力价值。它汇集了经济学文献的两股，关于政治家行动的决定因素和小额信贷解决的信贷市场失灵。这是第一个测试现有外部机构的差异如何影响小额贷款成功的研究，也是第一个提出和测试女性保留如何影响私人商品市场的具体模型的研究。用于女性保留的准随机选择过程使得数据分析中使用的识别策略直接且相当稳健，因此结果不能归因于反向或联合因果关系。该项目还可能对印度政府针对历史上受到歧视的群体（包括妇女）的保留地政策产生更广泛的影响。目前，印度政府为女性候选人保留了三分之一的选区；如果这个项目的结果表明女性领导者比男性领导者更善于解决信贷市场失灵问题，那么在考虑女性保留的最佳水平时，就应该考虑到这一信息。然而，在更概念化的层面上，这个项目的结果可能会挑战保留政策的目的；如果激励女性领导人解决这些信贷市场失灵的是政府补贴，而不是对其他女性的利他主义，那么如果受到适当政策的激励，就没有令人信服的理由说明男性不能做得像女性一样好。

This project investigates the interaction between microcredit groups (groups of low-income women who receive small, no-collateral, joint-liability loans) and local popular government in rural South India. Preliminary results from a limited dataset suggest that the gender of the local political leader has a strong impact on the quantity of credit received by microcredit groups in the leader's constituency; this is puzzling, because the leaders have no formal role in the provision of these loans. This project uses data gathered from an administrative block in Tamil Nadu, India to test different hypotheses about the mechanism by which leaders can increase lending and the motivation for why female leaders increase lending more than male leaders. The core of the project consists of two surveys, a survey of all group leaders and a survey of 5 randomly selected members from each group. The leader survey elicits information on the loans received by the group over the last 10 years; other general information on the group, such as the year of formation and the bank that the group is associated with; and a membership roster for the group. This roster forms the sample frame for the member survey, which elicits information on the household structure of the group member; information on the benefits her household has received through government schemes; and information on her relationship with local political leaders. These surveys are supplemented with election data for each term in each constituency, including data on which constituencies have been quasi-randomly reserved for female candidates by the government. Using this data, it is possible to test the hypothesis that female leaders increase lending in their constituencies by assuaging the bank's concern that the groups might default on their loans, and that they do this more than male leaders because it is cheaper for the bank to give female leaders a payoff due to the structure of loan subsidies in India. By testing to see if fluctuations in the rate of subsidization had differential effects on unsubsidized lending under male versus female leaders, it is possible to rule out the competing hypothesis that female leaders get more credit for their groups because they feel more altruism towards those groups than do male leaders. If unsubsidized lending under female leaders reacts more positively to increases in subsidies than unsubsidized lending under male leaders, this is evidence against the altruism hypothesis.This project has significant intellectual merit. It brings together two strands of the economics literature, on the determinants of action by politicians and on the credit market failures solved by microcredit. It is the first study to test how differnces in pre-existing outside institutions can affect the success of microlending, and the first study to present and test a specific model of how women's reservations can affect the market for a private good. The quasi-random selection process used for women's reservations makes the identification strategies used in the data analysis straightforward and quite robust, so that the results cannot be attributed to reverse or joint causality.This project could also have a broader impact on the Indian government's reservation policy towards groups historically subject to discrimination, including women. Currently, the Indian government reserves one-third of constituencies for female candidates; if the results of this project show that female leaders are better at resolving credit market failures than male leaders, that is information that should be taken into account while considering the optimal level of women's reservations. However, on a more conceptual level, the results of this project may challenge the purpose of the reservation policy; if female leaders are motivated to solve these credit market failures by government subsidies rather than by altruism towards other women, there is no compelling reason why men could not do as well as women if motivated by the proper policy.