Group versus Individual Liability: A Field Experiment in the Philippines

1
Group versus Individual LiabilityA Field
Experiment in the Philippines

• Xavier Giné
• World Bank

Dean Karlan Yale University Innovations for
Poverty Action
2
Motivation

• Microfinance is typically seen as a solution to
  credit market failures faced by the poor
• Group liability, a feature found in many micro
  loans is perceived as a key innovation that has
  contributed to this success

3
Motivation

• Yet, in recent years, many micro-lenders have
  expanded rapidly using individual liability
• In turn, this has motivated other lenders that
  were using group liability to shift to individual
  liability

4
Flexibility versus Sustainability

• (Perceived) tradeoff for deepening outreach of
  microfinance
• Flexibility (to increase demand)
• Price flexibility
• Liability flexibility
• Term flexibility
• Sustainability
• Repayment
• Covering fixed costs (reduce turnover)

5
Motivation

• Need for better understanding of relative merits
  of group versus individual liability lending.
• The best evidence would come from well-designed,
  deliberate experiments in which contracts are
  varied but everything else is kept the same
• The Economics of Microfinance
• Armendariz de Aghion and Morduch (2005)

6
Very Simple Summary

• What we did was very simple
• Took 169 preexisting centers of 25 clients in a
  group liability program
• Randomly chose 80 of them and removed the group
  liability.
• All else remained the same.
• Default did not change and more clients borrowed.

7
Theory and Evidence

• Literature focuses on repayment, despite being
  only one aspect of profitability
• Group lending solves informational asymmetries by
  shifting the burden from the lender to the
  clients (Ghatak and Guinnane, 1999)
• Adverse selection (screening and sorting)
• Moral hazard (ex-ante and ex-post)
• It also creates an element of insurance

8
Theory and Evidence

• Much less stress on client retention/growth
• Members may reject close friends for fear of
  social sanctions
• Some may be reluctant to borrow if information
  about other members is not available

9
Experimental Design

• Institution Green Bank of Caraga
• Rural Bank in the central Philippines
• Microfinance Operation started in 1999
• Group-liability lending program (BULAK) has over
  400 centers and 19,000 clients
• PAR (portfolio at risk) is 3.7

10
BULAK Program

• Program BULAK (Bangon Ug Lihok Alang sa
  Kalambuan)
• Target pop Business women in rural areas
• Two layers of liability group and center
• Initial Loan size P3000 - P5000 (60-90)
• Loan term between 4 - 6 months
• Weekly center meetings
• Weekly deposits in center, group, and personal
  savings accounts

11
Experimental Design

• Sample Framework
• 169 centers in Leyte Island (161 existing in Aug.
  2004 and 8 centers formed before Nov. 2004)
• Centers handled by 11 credit officers in 6
  branches
• Conversion of BULAK Centers
• Removing group liability
• Group members are no longer liable for each
  others loan.
• Dissolving center/group savings
• Savings are entirely personalized with NO
  change to total deposit.
• Randomization
• Stratified by credit officer and center age
• Done in three waves

12
Experimental Design
161 BULAK Centers
May 2005
August 2004
November 2004
January 2006
8 New centers opened
2 centers dissolved
4 centers dissolved
11 Centers Converted
24 Centers Converted
45 Centers Converted
Total 167 Centers 88 control 79
converted
Total 169 Centers 134 control 35
converted
150 control 11 converted
Total 135 Centers 66 control centers 69
converted centers
13
Results

• Default
• No change in default overall
• Better performance in converted centers for
  individuals with tighter social network
• Indicates that group liability is not necessary
  to mitigate moral hazard.
• Growth
• Retention increases
• Program attracts more clients
• Savings
• No change
• Loan size
• decreases, slightly (savings withdrawal effect?)

14
Results

• Evidence of monitoring effects
• Both baseline and new clients in converted
  centers remember less about other members
  defaults
• Evidence of selection effects
• New clients in converted centers are less likely
  to predict defaults of other members correctly.
• Social network
• Mostly no change. Some small evidence of fewer
  side-loans (insurance?) in converted centers.

15
Conclusion

• Evidence of mechanisms of screening monitoring.
• But they do not add up and lead to default!
• Why?
• Perhaps not enough time
• Perhaps simply not economically significant

16
Next Steps

• Design of New Areas (ongoing)

Converted to individual liability
Existing groups
Stay
Converted to individual liability after 1st
cycle
Group
Stay
New areas
Individual lending
Control