Group Lending and Enforcement October 2007 () Group lending October 2007 1 / 26
Group Lending in Theory Grameen I ("classic") 2:2:1 staggering at 4-6 week intervals 1 loan cycle = a year joint liability: formal sanctions in case of default initial small loan, growing with each loan cycle as credit history builds,! eventually large enough for house repairs, or sending child to university () Group lending October 2007 2 / 26
Mitigating Adverse Selection Can group lending make it possible to "implicitly" charge safe borrowers lower interest rates and keep them in the market? Joint liability ) incentive for "assortative matching" () Group lending October 2007 3 / 26
Example: 2 member group One-period project requiring $1 investment Bank s cost of $1 loan = k Fraction q of borrowers are "safe": gross return = y The remaining 1 q are "risky": ȳ with prob. p Gross return = 0 with prob. 1 p Indentical expected return: pȳ = y Borrowers know each others types, but lender doesn t Assortative matching ) a fraction q of groups are (safe, safe) () Group lending October 2007 4 / 26
If both types of borrower are in the market, what is the break-even repayment, ˆR b?,! assume that ȳ is large enough that ȳ > 2 ˆR b Then the probability of repayment by a risky pair is g = 1 (1 p) 2 = 2p p 2 > p since default occurs only if both members fail ) break even repayment: ˆR b = k q + (1 q)g This must be less than the minimum repayment without group lending R b = k q + (1 q)p () Group lending October 2007 5 / 26
Implications In this case risky borrowers can repay more often,! risk is transferred from bank to risky borrowers,! allows bank to lower interest rate and still break-even,! safe types may be lured back into the market () Group lending October 2007 6 / 26
Ex ante Moral Hazard and Joint Responsibility Based on Stiglitz (1990) Can group lending induce borrowers to provide required e ort? Joint liability ) incentive for members to impose sanctions on each other () Group lending October 2007 7 / 26
Example: 2 member group Projects require $1 investment Non-shirker generates output y for sure Shirker generates y with prob. p output = 0 with prob. 1 p Cost of providing e ort = c Gross interest rate = R Cost of funds to to lender = r () Group lending October 2007 8 / 26
Individual contract Borrower s IC constraint in individual contract: (y R) c p(y R) ) lender s maximum achievable lending rate R R = y c 1 p if R < r, this loan will not be made, even if y R > c () Group lending October 2007 9 / 26
Group contract Assumption: group members act to maximize expected group income,! any member that deviates can be "punished" y < 2r : if only one is successful, this is insu cient to cover sum of borrowing costs Borrowers IC constraint in group contract: (2y 2R) 2c p 2 (2y 2R) (y R) c p 2 (y R) ) lender s maximum achievable lending rate R = y c 1 p 2 > R If R > r > R, a shift to group lending contract allows this investment to go ahead () Group lending October 2007 10 / 26
Implications Group lending relaxes IC constraint ) more projects will be funded Idea can be extended to situations where internal group sanctions are costly () Group lending October 2007 11 / 26
Enforcement and Peer Monitoring Ex post moral hazard Recall our simple dynamic lending borrowing game with no saving,! discount factor δ,! borrower s output is F (L) where F 0 (L) > 0 and F 00 (L) < 0,! i = net opportunity cost of funds to lender Now allow possibility of peer monitoring in group lending ) each borrower must pay o debt of the other, if she reneges,! cost of monitoring = k,! probability of observing peer s output = q,! social sanction that can be applied to reneging borrower = d () Group lending October 2007 12 / 26
Individual Contract (benchmark) Borrower s incentive constraint: 1 δ [F (L) R] F (L) + 1 δ 1 δ v (IC),! lender s maximum feasible repayment: R R = δ [F (L) v] Suppose v is so high that R < (1 + i)l, for all values of L ) complete credit rationing () Group lending October 2007 13 / 26
Group borrowing contract Repayment only if 1 δ [F (L) R] F (L) q [R + d] + 1 δ 1 δ v (IC),! lender s maximum feasible repayment: R R = δ [F (L) v] + (1 δ)qd 1 (1 δ)q,! shifts IC constraint up Peer will monitor as long as expected gain exceed the cost,! introduces another constraint qr k () Group lending October 2007 14 / 26
R ZP IC 1 k/q IC 0 Figure: Enforcement Constraints under group lending L () Group lending October 2007 15 / 26
Implications Joint liability can make lending sustainable by inducing peer monitoring and overcoming the enforcement problem Relies heavily on use of "social sanctions",! is this realistic?,! is this a good thing? () Group lending October 2007 16 / 26
Group Lending in Practice BRAC in Bangladesh (Montgomery, 1996) (1) group lending works against most vulnerable individuals (2) village-level group plays key role in repayment, not 5-member group,! new borrowers may e ectively cover defaults of old Guatemala (Wydick, 1999),! social ties have little impact on repayment rates Thailand (Ahlin and Townsend, 2003),! in poorer regions (northeast) repayment rise with village level social sanctions,! in wealthier regions (central) default rates increase with extent of joint liability,! repayment rates decline with improvements in alternative borrowing sources () Group lending October 2007 17 / 26
FINCA in Peru (Karlan, 2003), Costa Rica (Wenner, 1995),! "social cohesion" matters for repayment rates,! default rates higher in wealthier towns Calmeadow in Toronto and Halifax (Gomez and Santor, 2003),! default less likely if members trust and/or know each other Philippines (Gine and Karlan, 2006),! compare individual to group lending in controlled experiment,! no impact on repayment rates () Group lending October 2007 18 / 26
Problems with Traditional Group Lending Mixed results across countries re ects di erences in trade-o between bene ts and costs Groups may be di cult/costly for borrowers to set up Attending group meetings can be costly in some cases; bene cial in others Transfers risk from bank to borrowers Beyond a certain lending scale, individual contracts may be preferred Collusion amongst borrowers () Group lending October 2007 19 / 26
Social sanctions for default often seem too harsh and/or not credible,! what if the defaulter has trouble through no fault of her own?,! punishment imposes a "deadweight loss",! Rai and Sjostrom (2004) propose "cross-reporting",! e ectively what often happens in practice Grameen II proposal,! "basic loan" (variable duration, seasonal varyiation in installments),! then " exible loan" (easier terms, but small) if borrower gets in trouble,! expulsion only if customer fails to repay this () Group lending October 2007 20 / 26
Beyond Group Lending Emerging view: joint liability is often not the main key to success Shift toward individual lending for the "not so poor" Emphasis on dynamic incentives to induce repayment,! e.g. proressive lending,! a key element of Grameen bank lending () Group lending October 2007 21 / 26
Progressive Lending (two period example) Invest L to get output y = AL > L Let δ = discount factor Let v = probability of being re nanced despite default gross borrowing rate = R IC constraint for xed L AL RL + δal AL + δval ) lender s maximum repayment: R = δa(1 v) () Group lending October 2007 22 / 26
Suppose loan grow by a factor λ > 1 between periods IC constraint AL RL + δλal AL + δval ) maximum repayment R = δa(λ v) > R If R < r < R, loans will be made that previously weren t () Group lending October 2007 23 / 26
Implications Progressive lending can help to relax borrowing constraint But, in a multi-period context, borrower may choose to default one loan size becomes large enough () Group lending October 2007 24 / 26
Why Frequent Repayment Installments? Creates an early warning system,! but why require repayment before investment pays o?,! and why impose this cost on the borrower? Provides lender with information about ability to repay independently of investment return Valuable for households that have di culty holding on to savings,! suggests that a key role of micro nance is to substitute for lack of savings institutions Fequency should depend on context () Group lending October 2007 25 / 26
Complementary Incentive Mechanisms Flexible approaches to collateral (i.e. not only land),! incentives to repay depend on threat of loss, not resale value to the lender Financial collateral,! works like a credit cooperative: micro lender also holds savings Public repayment,! uses threat of social stigma to induce repayment,! reduces opportunities for fraud Targetting of women,! empirical evidence suggests women are more reliable (not clear why) Cross reporting within groups,! information from another group member as to why a loan is not being repaid () Group lending October 2007 26 / 26