In a HKUST IEMS Academic Seminar, Sujata VIsaria (HKUST) examined the distributive impacts of two alternative approaches to deliver agricultural credit to smallholders: TRAIL (or trader-agent intermediated lending), where local traders recommend village residents for individual liability micro-loans, and GBL (or group-based lending), where households self-select into groups and receive joint liability loans. This is joint work with Pushkar Maitra, Sandip Mitra and Dilip Mookherjee. Find out more about the event at http://iems.ust.hk/trail
Sujata Visaria: Finding the Diamonds in the Rough - Welfare Comparisons of Two Targeting Mechanisms in Microcredit
1. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Finding the Diamonds in the Rough
The Equity Impacts of Targeting Smallholder Agricultural Credit
Pushkar Maitra, Sandip Mitra, Dilip Mookherjee,
Sujata Visaria
HKUST Institute for Emerging Market Studies
9 March 2017
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 1 / 50
2. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Targeting Private Benefits
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 2 / 50
3. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Targeting Private Benefits
How should governments choose the recipients of private
benefits?
subsidies: food, fertilizer
transfer payments: pensions
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 2 / 50
4. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Targeting Private Benefits
How should governments choose the recipients of private
benefits?
subsidies: food, fertilizer
transfer payments: pensions
Centralized schemes: lack of information
Decentralized schemes: corruption & elite capture
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 2 / 50
5. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Smallholder Agricultural Credit
Designing agricultural credit programs for small farmers
State-sponsored efforts: poor information leads to high default
Microcredit: high repayment, but does not usually target
agriculture
Can we design a scheme that
leverages local information? and
avoids problems of elite capture?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 3 / 50
6. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Agent-intermediated Lending
MMMMV (2017) designed an alternative credit delivery
mechanism (AIL):
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 4 / 50
7. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Agent-intermediated Lending
MMMMV (2017) designed an alternative credit delivery
mechanism (AIL):
Leverages local information, but differently
Individual liability loans
Borrowers recommended by a local “agent” embedded in the
community
Agent incentivized by commissions that depend on repayment
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 4 / 50
8. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Agent-intermediated Lending
MMMMV (2017) designed an alternative credit delivery
mechanism (AIL):
Leverages local information, but differently
Individual liability loans
Borrowers recommended by a local “agent” embedded in the
community
Agent incentivized by commissions that depend on repayment
Field experiment in 48 villages
TRAIL loans had positive significant average impacts on farm
value-added: 22%
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 4 / 50
9. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Average Treatment Effects: Farm and Non-Farm
Income (Efficiency)
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 5 / 50
10. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
The Mechanism
TRAIL agents are more likely to select “able” borrowers
high productivity
low risk of crop failure
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 6 / 50
11. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
The Mechanism
TRAIL agents are more likely to select “able” borrowers
high productivity
low risk of crop failure
Positive impacts are larger for the more able farmers
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 6 / 50
12. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
The Comparison: Group-based Lending
Group-based lending involves self-selected borrower groups
Joint liability lending
Participation does not require recommendation by an agent
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 7 / 50
13. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
The Comparison: Group-based Lending
Group-based lending involves self-selected borrower groups
Joint liability lending
Participation does not require recommendation by an agent
Both high and low ability individuals will form groups
Because both pay lower interest rates on GBL loans than the
status quo
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 7 / 50
14. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
The Comparison: Group-based Lending
Group-based lending involves self-selected borrower groups
Joint liability lending
Participation does not require recommendation by an agent
Both high and low ability individuals will form groups
Because both pay lower interest rates on GBL loans than the
status quo
GBL loans had non-significant average impacts on farm
value-added
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 7 / 50
15. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Average Treatment Effects: Farm and Non-Farm
Income (Efficiency)
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 8 / 50
16. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Research Question: Equity Impacts
Previous work examines the relative efficiency impacts
Welfare impacts depend on both efficiency and equity impacts
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 9 / 50
17. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Research Question: Equity Impacts
Previous work examines the relative efficiency impacts
Welfare impacts depend on both efficiency and equity impacts
Could GBL dominate TRAIL on equity grounds?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 9 / 50
18. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Why Worry?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 10 / 50
19. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Why Worry?
MMMMV (2017): 30-40% of the efficiency impact of TRAIL
due to borrower selection:
TRAIL agents selected borrowers who were more productive
Productive borrowers may also be wealthier/higher SES to
begin with
TRAIL could amplify these differences
Efficiency gains, but at the cost of equity?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 10 / 50
20. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Aggregate Welfare Impacts
We estimate the scheme’s impact on aggregate welfare
depends on both efficiency and equity
By partitioning households
by landholding
by social group: religion/caste
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 11 / 50
21. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 12 / 50
22. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts
Question: If the scheme was introduced into a village, by how
much would the welfare of representative household change?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 12 / 50
23. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts
Question: If the scheme was introduced into a village, by how
much would the welfare of representative household change?
Problem: selection into the scheme was NOT random
Average TRAIL borrower NOT representative of population
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 12 / 50
24. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts, contd.
Question: If the scheme was introduced into a village, by how
much would the income of representative household change?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 13 / 50
25. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts, contd.
Question: If the scheme was introduced into a village, by how
much would the income of representative household change?
That depends on:
does that household participate in the scheme (pg )?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 13 / 50
26. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts, contd.
Question: If the scheme was introduced into a village, by how
much would the income of representative household change?
That depends on:
does that household participate in the scheme (pg )?
if it participates, how much does its income change (CTEg )?
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 13 / 50
27. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts, contd.
Question: If the scheme was introduced into a village, by how
much would the income of representative household change?
That depends on:
does that household participate in the scheme (pg )?
if it participates, how much does its income change (CTEg )?
Expected income effect for group g = pg × CTEg
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 13 / 50
28. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
How we estimate welfare impacts, contd.
Question: If the scheme was introduced into a village, by how
much would the income of representative household change?
That depends on:
does that household participate in the scheme (pg )?
if it participates, how much does its income change (CTEg )?
Expected income effect for group g = pg × CTEg
Use welfare functions to infer welfare effects
with different assumptions about inequality aversion
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 13 / 50
29. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
TRAIL Features
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 14 / 50
30. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
TRAIL Features
Agent is selected randomly from list of established
trader/lenders in any given village
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 14 / 50
31. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
TRAIL Features
Agent is selected randomly from list of established
trader/lenders in any given village
Agent recommends 30 borrowers from among landless or
marginal landowners (≤ 1.5 acres)
subset of these are chosen randomly to receive offer of TRAIL
loans
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 14 / 50
32. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
TRAIL Features
Agent is selected randomly from list of established
trader/lenders in any given village
Agent recommends 30 borrowers from among landless or
marginal landowners (≤ 1.5 acres)
subset of these are chosen randomly to receive offer of TRAIL
loans
Agent plays no further role: MFI sets loan terms, directly
lends to and collects repayments from borrowers
No group meetings, savings requirements or gender
restrictions
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 14 / 50
33. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
TRAIL Loan Features
Loan interest rate pegged at below average rates on informal
credit in the village
Dynamic borrower incentives: start with small loans, future
credit access grows at fast rate based on current repayment
*Loan durations/timing: 4 months, match key-crop cycles
*Insurance against covariate (price-yield) risks
*Doorstep banking, no bank accounts
(*: non-standard)
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 15 / 50
34. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
The Group Based Lending Scheme
Same loan features as TRAIL loans, except:
5-member self-formed groups of borrowers
joint liability
monthly group meetings
savings requirements
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 16 / 50
35. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Experimental Setting
Two potato-growing districts in a state in eastern India
48 villages randomly assigned to:
TRAIL scheme: 24
GBL scheme: 24
Collaborated with a microfinance institution
8 4-month cycles: Sept 2010 - July 2013
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 17 / 50
36. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Partitions
Land class
Landless
0-0.5 acres
0.5-1 acre
1-1.5 acres
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 18 / 50
37. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Partitions
Land class
Landless
0-0.5 acres
0.5-1 acre
1-1.5 acres
Social group
scheduled castes/scheduled tribes
non-Hindus
general caste Hindus
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 18 / 50
39. HH Characteristics by Social Group
Total SC/ST Non-Hindu General
(1) (2) (3) (4)
Head’s education: primary school or more 0.420 0.222 0.317 0.614
(0.011) (0.015) (0.025) (0.016)
Head’s primary occupation: cultivation 0.439 0.284 0.432 0.560
(0.011) (0.017) (0.026) (0.016)
Head?s primary occupation: casual labor 0.331 0.562 0.276 0.176
(0.010) (0.018) (0.024) (0.012)
Landholding (acres) 0.465 0.293 0.482 0.590
(0.009) (0.013) (0.022) (0.014)
Pucca house 0.273 0.112 0.329 0.376
(0.010) (0.012) (0.025) (0.016)
Electrified house 0.750 0.604 0.792 0.846
(0.010) (0.018) (0.022) (0.012)
Own a radio/ TV/ VCR/ DVD 0.453 0.318 0.387 0.583
(0.011) (0.017) (0.026) (0.016)
Owns motorized vehicle 0.126 0.030 0.155 0.189
(0.007) (0.006) (0.019) (0.013)
Owns phone (mobile or landline) 0.590 0.413 0.691 0.687
(0.011) (0.018) (0.025) (0.015)
Has savings bank account 0.446 0.284 0.408 0.586
(0.011) (0.017) (0.026) (0.016)
Has insurance policy 0.287 0.161 0.237 0.403
(0.010) (0.013) (0.023) (0.016)
40. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Land Class
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 21 / 50
41. Step 1: Selection Frequencies pg
pg =
number of g hhs selected
total number of g hhs
selected =
recommended by TRAIL agents
self-selected into GBL groups
selection = take-up
only households with ≤ 1.5 acres of land considered
42. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 1: Selection Frequencies, contd.
0.059
0.073 0.074
0.054
0.068
0.040
0.070
0.032
0.05.1
Landless Land 0-0.5 Land 0.5-1.0 Land 1.0-1.5
TRAIL GBL
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 23 / 50
43. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Selection Frequencies by Land Class
the landless have a slightly lower selection frequency in
TRAIL than in GBL
TRAIL: across land classes, selection frequencies are very
similar
GBL: selection frequencies fall as landholding increases
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 24 / 50
44. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 2: CTEs, by land
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 25 / 50
45. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 2: CTEs, by land
-20000020000
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
TRAIL, absolute
-20000020000
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
GBL, absolute
Conditional Treatment Effects, by Land
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 25 / 50
46. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Absolute v. Proportional CTEs, by land: TRAIL
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 26 / 50
47. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Absolute v. Proportional CTEs, by land: TRAIL
01000020000
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
TRAIL, absolute
-1012
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
TRAIL, relative
Conditional Treatment Effects, by Land
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 26 / 50
48. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Absolute v. Proportional CTEs, by land: GBL
-20000-10000010000
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
GBL, absolute
-.50.511.5
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
GBL, relative
Conditional Treatment Effects, by Land
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 27 / 50
49. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 3: Expected treatment effects on income
-.010.01.02.03.04.05
ty_g
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
TRAIL, relative
-.010.01.02.03.04.05
ty_g
Landless
0-0.5 acres
0.5-1 acres
1-1.5 acres
GBL, relative
Expected Proportional Treatment Effects, by Land
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 28 / 50
50. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 0
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 29 / 50
51. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 0
-10123
rdw
TRAIL GBL
theta=0
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 29 / 50
52. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 1
0.05.1.15.2
rdw
TRAIL GBL
theta=1
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 30 / 50
53. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 2
20040060080010001200
rdw
TRAIL GBL
theta=2
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 31 / 50
54. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 3
05000001.0e+061.5e+062.0e+06
rdw
TRAIL GBL
theta=3
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 32 / 50
55. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 4
01.0e+092.0e+093.0e+09
rdw
TRAIL GBL
theta=4
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 33 / 50
56. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 5
01.0e+122.0e+123.0e+124.0e+12
rdw
TRAIL GBL
theta=5
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 34 / 50
57. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Social Group
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 35 / 50
58. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 1: Selection frequencies
0.073 0.073
0.066
0.050
0.064
0.037
0.05.1
SC/ST Non-Hindu General
Probabilities of selection
TRAIL GBL
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 36 / 50
59. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 2: CTEs, by social group
-500005000
SC/ST
Non-Hindu
General
TRAIL, absolute
-500005000
SC/ST
Non-Hindu
General
GBL, absolute
Conditional Treatment Effects, by Social Group
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 37 / 50
60. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Absolute v. Proportional CTEs, by social group:
TRAIL
0500010000
SC/ST
Non-Hindu
General
TRAIL, absolute
0.51
SC/ST
Non-Hindu
General
TRAIL, relative
Conditional Treatment Effects, by Social Group
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 38 / 50
61. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Absolute v. Proportional CTEs, by social group:
GBL
-500005000
SC/ST
Non-Hindu
General
GBL, absolute
-.50.5
SC/ST
Non-Hindu
General
GBL, relative
Conditional Treatment Effects, by Social Group
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 39 / 50
62. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 3: Expected treatment effects, by social
group
-.010.01.02.03.04.05
ty_g
SC/ST
Non-Hindu
General
TRAIL, relative
-.010.01.02.03.04.05
ty_g
SC/ST
Non-Hindu
General
GBL, relative
Expected Proportional Treatment Effects, by Social Group
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 40 / 50
63. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 4: Change in welfare, θ = 0
0.005.01.015.02.025
rdw
TRAIL GBL
theta=0
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 41 / 50
64. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 3: Change in welfare, θ = 4
5.0e+061.0e+071.5e+072.0e+07
rdw
TRAIL GBL
theta=4
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 42 / 50
65. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Step 3: Change in welfare, θ = 5
1.0e+092.0e+093.0e+094.0e+095.0e+096.0e+09
rdw
TRAIL GBL
theta=5
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 43 / 50
67. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Discussion
The evidence that TRAIL was elitist is limited.
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 45 / 50
68. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Discussion
The evidence that TRAIL was elitist is limited.
Comparing landless borrowers:
in TRAIL had lower baseline farm income than in GBL
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 45 / 50
69. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Discussion
The evidence that TRAIL was elitist is limited.
Comparing landless borrowers:
in TRAIL had lower baseline farm income than in GBL
In fact, GBL might be keeping out
both the high wealth/high status households
and the poorest among the low wealth households
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 45 / 50
71. Social Capital in GBL Groups
Total SC/ST Non-Hindu General
(1) (2) (3) (4)
Invited by group members 0.583 0.536 0.800 0.558
(0.032) (0.048) (0.069) (0.051)
Encouraged by officials to join group 0.417 0.464 0.200 0.442
(0.032) (0.048) (0.069) (0.051)
Received assistance from others in the group (Cycle 1) 0.248 0.239 0.361 0.216
(0.028) (0.040) (0.081) (0.042)
Provided assistance to others in the group (Cycle 1) 0.124 0.080 0.171 0.158
(0.021) (0.026) (0.065) (0.038)
Proportion members with same occupation 0.624 0.616 0.598 0.642
(0.015) (0.022) (0.036) (0.023)
Proportion members with same education level 0.530 0.520 0.594 0.516
(0.012) (0.019) (0.034) (0.018)
Proportion members in same social group 0.522 0.732 0.887 0.192
(0.027) (0.038) (0.044) (0.033)
Herfindahl Index by social group 0.762 0.873 0.940 0.593
(0.014) (0.019) (0.024) (0.019)
Coefficient of variation of landholding 0.999 1.217 0.975 0.805
(0.035) (0.060) (0.075) (0.044)
72. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Summary of Results
Compare welfare impacts of two alternative credit delivery
models
TRAIL is not more elitist than GBL
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 48 / 50
73. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Summary of Results
Compare welfare impacts of two alternative credit delivery
models
TRAIL is not more elitist than GBL
When partitioning by landholding:
TRAIL generates higher average gains and benefits the
landless more than GBL
As a result it dominates GBL.
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 48 / 50
74. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Summary of Results
Compare welfare impacts of two alternative credit delivery
models
TRAIL is not more elitist than GBL
When partitioning by landholding:
TRAIL generates higher average gains and benefits the
landless more than GBL
As a result it dominates GBL.
When partitioning by social group:
TRAIL: High status group outperforms other groups
GBL: Low status groups outperform high status groups
Point estimates indicate larger welfare in GBL under high θ
But differences are not statistically significant
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 48 / 50
75. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Conclusion
Our previous work indicates that TRAIL is superior to GBL
from an efficiency standpoint.
From an equity standpoint:
Clear evidence from land partitioning that it outperforms GBL
Less clear evidence from social group partitioning
differences depend on θ values
lack statistical significance
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 49 / 50
76. Equity Impacts of
Targeting Credit
MMMV
Motivation
Equity Impacts
TRAIL & GBL
Partitions
Results
Land
Social Group
Inequality
Mechanisms
Conclusion
Thank you!
MMMV (Mar 2017) Equity Impacts of Targeting Credit Mar 2017 50 / 50