Strategic Default in joint liability groups: Evidence from a natural experiment in India

Transcription

1 Strategic Default in joint liability groups: Evidence from a natural experiment in India Xavier Gine World Bank Karuna Krishnaswamy IFMR Alejandro Ponce World Justice Project CIRANO, November 9-10,

2 Motivation Microfinance is typically seen as a solution to credit market failures faced by the poor Among the different features of micro loans (escalating loan sizes, frequency of repayment, etc), group lending is perceived as a key innovation driving the high repayment rates claimed by programs around the world Group lending contracts often include a joint liability clause but this is not always the case 2

3 Motivation The literature points out several advantages of joint liability: It solves informational asymmetries by shifting the burden from the lender to the clients (Ghatak and Guinnane, 1999) resulting in lower transaction costs for the institution. Adverse selection (screening and sorting) Moral hazard (ex-ante and ex-post) As well as several disadvantages: Attendance in group meetings and monitoring can be costly. If a critical number of group members cannot repay, members that would otherwise repay find it in their interest not to (Besley and Coate, 1995). 3

4 Motivation More generally, while some groups do default and disband, we know little about why repayment deteriorates. Exposure to common shocks Strategic reasons (Besley and Coate, 1995) Related to Breza (2012) and Allen (2012) Understanding the actual mechanism is important for the design of microcredit products that minimize the occurrence of group default. 4

5 This Paper Provide evidence of strategic default using a natural experiment in India. We ask: Are members of a joint liability group more likely to default when the fraction of defaulters in the group rises? MFIs in the state of Karnataka, India, operate through group-lending schemes, with borrowers sorting into religiously-mixed groups according to their location. On Jan 28 th, 2009, the Anjuman Committee of Kolar issued a statement banning all Muslims from repaying their MFI loans claiming that charging interest was haram As a result, borrowers in Muslim-dominated groups faced, after the announcement, a greater repayment burden compared to borrowers in Hindu-dominated groups. 5

6 Default rate at maturity muslim hindu

7 Default rate at maturity muslim hindu

8 Default rate at maturity muslim hindu

9 Default rate (Muslims in mixed centers) high low

10 Default rate (Hindus in mixed centers) 0.6 high low

11 Institutional setting Clients are predominantly women, with Muslims representing 53% of MFI clients. Loans are transacted and recorded on an individual basis, but an outstanding default by any group member renders the entire group ineligible for future loans. Loans range between Rs 500 and Rs 30,000, depending on the loan cycle. MFIs do not collect savings that could be used as collateral. Al least 12% of MFI borrowers in Kolar and Ramanagaram had more than one outstanding loan in an MFI during (No credit bureau). 11

12 Data MIS from 7 out of 8 registered MFIs operating in Kolar and Ramanagaram. All loans given from 2007 (for 4 MFIs) and 2008 (for 3 MFIs) until Dec 2009 Loan Characteristics and repayment performance at maturity Basic information on borrowers: name, father name, religion, etc Phonetic algorithm to match customers across the databases of multiple MFIs We only use loans disbursed before Anjuman announcement Survey from 600 MFI clients in Kolar and Ramanagaram 12

13 Identification Strategy: Intuition The fatwa provides an exogenous repayment burden for borrowers in Muslim-dominated groups Key Question: Do borrowers in Muslim-dominated centers default more after the announcement compared to borrowers in Hindu-dominated groups? One possible concern is that borrowers in different groups are inherently different. To address this, we focus on a sample of borrowers with multiple loans (in different centers) Identification comes from variation in the behavior of the same individual across different groups. 13

14 Identification Strategy: Checks Joint liability at the center level should be enforced: Check for a drop in loans disbursed after announcement, following increase in defaults 14

15 Number of loans by date of disbursement Muslims Hindus

16 Number of loans by date of disbursement (Hindus) Hindus in center with at least one defaulter Hindus in center without any defaulter 16

17 Identification Strategy: Checks Joint liability at the center level should be enforced: Check for a drop in loans disbursed after announcement, following increase in defaults There should be no reduction in loan sizes The percentage of borrowers in good standing in centers in which at least one member defaulted among borrowers in those centers is 93/21,039 = 0.4% 17

18 Evidence of strategic default Distribution of center and colony-level defaults

19 Pct of members in default in center Evidence of strategic default (center) Pct of muslims in center 0 (0,20] (20,40] (40,60] (60,80] (80,100) 100 Total 100 4% 12% 28% 30% 55% 47% 80% 61% (80,100) 4% 11% 16% 13% 6% 14% 6% 7% (60,80] 1% 5% 6% 13% 19% 8% 2% 4% (40,60] 2% 6% 6% 4% 4% 13% 2% 3% (20,40] 4% 6% 3% 0% 4% 6% 2% 3% (0,20] 5% 9% 0% 9% 0% 5% 1% 2% 0 79% 51% 41% 30% 11% 7% 7% 20% Total 100% 100% 100% 100% 100% 100% 100% 100% Pct with 0% or 100% default Pct centers by density of muslims 83% 63% 69% 61% 66% 54% 87% 81% 13% 5% 2% 2% 4% 8% 66% 100%

20 Pct of members in default in center Evidence of strategic default (center) Pct of muslims in center 0 (0,20] (20,40] (40,60] (60,80] (80,100) 100 Total 100 4% 12% 28% 30% 55% 47% 80% 61% (80,100) 4% 11% 16% 13% 6% 14% 6% 7% (60,80] 1% 5% 6% 13% 19% 8% 2% 4% (40,60] 2% 6% 6% 4% 4% 13% 2% 3% (20,40] 4% 6% 3% 0% 4% 6% 2% 3% (0,20] 5% 9% 0% 9% 0% 5% 1% 2% 0 79% 51% 41% 30% 11% 7% 7% 20% Total 100% 100% 100% 100% 100% 100% 100% 100% Pct with 0% or 100% default Pct centers by density of muslims 83% 63% 69% 61% 66% 54% 87% 81% 13% 5% 2% 2% 4% 8% 66% 100%

21 Pct of members in default in center Evidence of strategic default (center) Pct of muslims in center 0 (0,20] (20,40] (40,60] (60,80] (80,100) 100 Total 100 4% 12% 28% 30% 55% 47% 80% 61% (80,100) 4% 11% 16% 13% 6% 14% 6% 7% (60,80] 1% 5% 6% 13% 19% 8% 2% 4% (40,60] 2% 6% 6% 4% 4% 13% 2% 3% (20,40] 4% 6% 3% 0% 4% 6% 2% 3% (0,20] 5% 9% 0% 9% 0% 5% 1% 2% 0 79% 51% 41% 30% 11% 7% 7% 20% Total 100% 100% 100% 100% 100% 100% 100% 100% Pct with 0% or 100% default Pct centers by density of muslims 83% 63% 69% 61% 66% 54% 87% 81% 13% 5% 2% 2% 4% 8% 66% 100%

22 Pct of members in default in colony Evidence of strategic default (Colony) Pct of muslims in colony 0 (0,20] (20,40] (40,60] (60,80] (80,100) 100 Total 100 0% 0% 0% 0% 0% 0% 14% 4% (80,100) 0% 0% 0% 0% 0% 50% 0% 16% (60,80] 0% 0% 0% 50% 67% 29% 24% 21% (40,60] 0% 11% 50% 25% 0% 17% 19% 17% (20,40] 10% 22% 17% 25% 33% 4% 29% 17% (0,20] 0% 22% 33% 0% 0% 0% 0% 5% 0 90% 44% 0% 0% 0% 0% 14% 21% Total 100% 100% 100% 100% 100% 100% 100% 100% Pct with 0% or 100% default Pct colonies by density of muslims 90% 44% 0% 0% 0% 0% 29% 25% 11% 12% 8% 5% 4% 32% 27% 100%

23 Evidence of strategic default Distribution of center and colony-level defaults Evidence from the survey

24 Evidence of strategic default Distribution of center and colony-level defaults Evidence from the survey Muslim Borrowers Borrowers Borrowers in centers in mixed with only centers Muslims Hindu Borrowers Borrowers in centers with only Hindus Borrowers in mixed centers Number of borrowers Number of loans Pct. of loans that defaulted after the fatwa 67% 55% 2% 25% Since January 2009, did any center member borrow from you to make a repayment? (1=Yes) Pct. responding "I had the money, but I did not want to repay" if missed payment on a loan 53% 44% 93% 63% 100% 97% 33% 100%

25 Regression specification (D in D) For loan l of individual i in center c at time t: Y lict = ap t + bs c + d P t x S c + D ic + ε lict Y lict = outcome of interest P t = Post announcement indicator S c = Share of Muslims in center D ic = Individual fixed effect Standard errors clustered at center level

26 Variables 2 outcomes: Dummy of default at maturity; Balance outstanding at maturity; Focus on outcomes at maturity We do not have loan performance post maturity, but credit officers report that almost no loans were repaid post maturity Defaults are recorded on an individual basis (but joint liability is enforced on a group basis). In our data, it is not always the case that all members of the group default

27 Loan in Default at Maturity All Borrowers Borrowers with loans from multiple centers (1) (2) (3) (4) Post 0.169*** 0.214*** 0.262*** 0.216*** (0.027) (0.040) (0.046) (0.050) Pct Muslims in Center 0.001*** 0.001*** (0.000) (0.000) (0.001) (0.001) Pct Muslims in Center x Post 0.006*** 0.005*** 0.005*** 0.004*** (0.000) (0.000) (0.001) (0.001) Pct Muslims in Colony x Post (0.001) Mean Dep. Var Number of Loans 35,012 10,313 10,313 10,313 Number of Borrowers 19,888 2,975 2,975 2,975 Number of Centers 1,376 1,096 1,096 1,096 R-squared Borrower Fixed Effects No No Yes Yes 27

28 Loan in Default at Maturity All Borrowers Borrowers with loans from multiple centers (1) (2) (3) (4) Post 0.169*** 0.214*** 0.262*** 0.216*** (0.027) (0.040) (0.046) (0.050) Pct Muslims in Center 0.001*** 0.001*** (0.000) (0.000) (0.001) (0.001) Pct Muslims in Center x Post 0.006*** 0.005*** 0.005*** 0.004*** (0.000) (0.000) (0.001) (0.001) Pct Muslims in Colony x Post (0.001) Mean Dep. Var Number of Loans 35,012 10,313 10,313 10,313 Number of Borrowers 19,888 2,975 2,975 2,975 Number of Centers 1,376 1,096 1,096 1,096 R-squared Borrower Fixed Effects No No Yes Yes 28

29 Loan in Default at Maturity All Hindus Hindu borrowers with loans from multiple centers (1) (2) (3) (4) Post 0.159*** 0.204*** 0.239*** (0.027) (0.041) (0.046) (0.047) Pct Muslims in Center 0.001** (0.000) (0.000) (0.001) (0.001) Pct Muslims in Center x Post 0.005*** 0.006*** 0.006*** 0.004*** (0.001) (0.001) (0.001) (0.001) Pct Muslims in Colony x Post 0.007*** (0.001) Mean Dep. Var Number of Loans 9,449 3,106 3,106 3,106 Number of Borrowers 4, Number of Centers R-squared Borrower Fixed Effects No No Yes Yes 29

30 Alternative Interpretations Our exercise identified large differences in default rates for borrowers in Muslim-dominated centers relative to borrowers in Hindu-dominated groups. But, what are the precise mechanisms through which this effect takes place? Differential debt burden after fatwa?

31 Differences in Debt Burden All Hindus Hindus belonging to at least one mixed center Hindus with loans in multiple centers Hindus with loans in multiple centers, with at least one being mixed All Muslims Muslims belonging to at least one mixed center Muslims with loans in multiple centers Muslims with loans in multiple centers, with at least one being mixed Fixed effects? Loan size by Number of Interest rate maturity installments (1) (2) (3) No *** 0.055** [5.375] [0.000] [0.026] Yes [14.034] [0.000] [0.023] Yes * 0.076*** [11.357] [0.000] [0.027] Yes ** [11.387] [0.000] [0.030] No [6.276] [0.000] [0.020] Yes * [11.475] [0.000] [0.021] Yes [11.759] [0.000] [0.035] Yes [12.087] [0.000] [0.037] 31

32 Alternative Interpretations Our exercise identified large differences in default rates for borrowers in Muslim-dominated centers relative to borrowers in Hindu-dominated groups. But, what are the precise mechanisms through which this effect takes place? Differential debt burden after fatwa? No! Borrowers in muslim-dominated groups are different or are exposed to more pressure not to repay.

33 Alternative Interpretations Our exercise identified large differences in default rates for borrowers in Muslim-dominated centers relative to borrowers in Hindu-dominated groups. But, what are the precise mechanisms through which this effect takes place? Differential debt burden after fatwa? No! Borrowers in muslim-dominated groups are different? No! Borrowers in muslim-dominated groups face higher transactions costs to repay?

34 Alternative Interpretations Our exercise identified large differences in default rates for borrowers in Muslim-dominated centers relative to borrowers in Hindu-dominated groups. But, what are the precise mechanisms through which this effect takes place? Differential debt burden after fatwa? No! Borrowers in muslim-dominated groups are different? No! Borrowers in muslim-dominated groups face higher transactions costs to repay? No! Borrowers in muslim-dominated groups face less attractive loans in future?

35 Average loan size by date of disbursement Muslims Hindus

36 Alternative Interpretations Our exercise identified large differences in default rates for borrowers in Muslim-dominated centers relative to borrowers in Hindu-dominated groups. But, what are the precise mechanisms through which this effect takes place? Differential debt burden after fatwa? No! Borrowers in muslim-dominated groups are different or are exposed to more pressure not to repay? No! Borrowers in muslim-dominated groups face higher transactions costs to repay? No! Borrowers in muslim-dominated groups face less attractive loans in future? No!

37 Still to be done Evidence of strategic default in Grameen group liability groups triggered conversion to invididual liability under Grameen II (Dowla and Barua, 2006) What are the welfare implications of removing joint liability? Develop a choice model of the repayment decision and estimate it Use structure of model to evaluate different liability structures 37