2. Motivation
Castilla & Walker (2013):
Ghana
Private lottery windfalls are allocated towards gifts and loans
Public lottery windfalls are spent on highly visible items (assets,
ceremonies and health) – Husbands
Personal care by wives
Castilla (2011):
Ghana (same households, but 10 years earlier)
Asymmetric Information over farm income allocated towards
gifts by husbands.
Theory: incentives to hide in HH with cooperative and noncooperative contracts.
3. Research Objectives
Bargaining power depends on:
Share of resources of each spouse
Anderson and Eswaran, (2009): the influence of additional
resources on bargaining power of women is small compared to
idiosyncratic or cultural sources of bargaining power.
Cultural norms and socio-economic characteristics.
Research Question:
How
do different sources of bargaining power affect
cooperation and efficiency between spouses under asymmetric
information?
4. Conclusions
Theoretical Findings:
There exists a threshold level change of bargaining power that
needs to be exceeded to induce revelation
Empirical Findings:
Experiments with established couples in Uttarakhand, India
Ultimatum games that varied control over endowments
Information Treatments: Public, Private and Private with option
to reveal
When given the chance, 21% of spouses choose to conceal
No differences by gender contributions towards HH account
Men who choose to reveal are more cooperative than women
5. Experiments with Spouses
Testing across HH models: Kebede et al. (2011); Munro et
al. (2008); Munro et al. (2011)
Spousal Preferences: Bateman & Munro (2005); Cochard et
al. (2009); Dasgupta and Mani, (2013); Carlsson et al.
(2012)
Dictator games: Iversen et al. (2011)
Asymmetric Information: Ashraf (2009); Mani (2013);
Hoel (2013)
6. Experiments with Spouses
Ashraf (2009) the goal is to find whether hiding occurs,
and how communication influences spousal behavior.
Information Treatments:
Private: obscures choices and allows plausible deniability
Public: all choices and outcomes are observed
Negotiation: spouses communicate prior to making choices
Findings:
Invest in own account if Private (men and women)
Commit consumption if Public (men)
Save in wife’s account if Negotiation (men)
If spouse controls savings, save in own account
7. Experiments with Spouses
Mani (2013): differs in timing of information treatment
Income-control treatment:
Information Treatment: (ex-post)
Two investment options (high and low)
Baseline: Fixed shares, no trade-off btw control and payoff
High return investment goes towards spouse’s account: private,
cash or joint – differences in cost of appropriating
Low return: own private account
None, Full or partial – only inform outcomes (not choices) and
allow to renegotiate
Findings: control matters, information and threat of
retaliation do not.
8. The Model
Contract:
Spouses pool all of their resources and bargain over how to allocate
them between private and public good consumption.
Benchmark: When T is revealed:
The household solves:
max
𝑄,𝑥 𝑚 ≥0
𝜇 𝑌𝑓 , 𝑌 𝑚 , 𝑇, 𝑝, 𝑧
+ 1 − 𝜇 𝑌𝑓 , 𝑌 𝑚 , 𝑇, 𝑝, 𝑧
𝑢 𝑥𝑚 + 𝑣 𝑄
𝑢 𝑌𝑓 + 𝑌 𝑚 + 𝑇 − 𝑥 𝑚 − 𝑝𝑄 + 𝑣 𝑄
Where:
𝜇 𝑌𝑓 , 𝑌 𝑚 , 𝑇, 𝑝, 𝑧
− 𝜇 𝑌𝑓 , 𝑌 𝑚 , 𝑇,z 𝑝, 𝑧
𝑢 𝑥 𝑚 Husband’s bargaining power
+ 𝑣 𝑄
𝑢 Distribution𝑇factors, for instance𝑄 cultural or legal factors
𝑌𝑓 + 𝑌 𝑚 + − 𝑥 𝑚 − 𝑝𝑄 + 𝑣
9. The Model
Proposition:
There exists a strictly positive threshold change in bargaining
power (∆𝜇 ) such that if 𝜕𝜇 < ∆𝜇 there are incentives to hide.
𝜕𝑇
Intuition of Proof: Husband hides T iff
𝜕𝑈 𝑚
𝜕𝑇
𝑅
<
𝜕𝑈 𝑚
𝜕𝑇
𝐻
Low Initial
Bargaining Power
lim ∆𝜇 < 0
No Hiding
High Initial
Bargaining Power
lim ∆𝜇 > 0
Hiding can occur
depending on
responsiveness of BP
𝜇 →0
𝜇 →1
10. Testable Hypotheses
Hypothesis 1: Hiding is Inefficient.
Hiding will cause underinvestment in household public goods.
Hypothesis 2: Income hiding can occur even when revealing the
available resources increases bargaining power.
Bargaining power is determined partly by relative amount of
resources and partly by other factors.
12. Experimental Design
Sample: 200 established couples, 50% from Dehradun and
50% from Almora.
Recruitment: Door-to-door
Mobile Lab: at their homes.
Incentives: LED lamp per spouse, plus opportunity to earn
some cash.
Protocol:
Enumerator of same gender
Separate rooms
Ultimatum Game: using strategy method
Survey
13. Experimental Design: Timeline
Randomly assigned to an information treatment
Each spouse randomly assigned to a Role (A or B)
Player A: Informed party
Player B: Informed of outcome
Endowments: informed of their own (ω_i ) and their
spouse’s (ω_j ) endowment
Each spouse was presented with 7 decisions, one-at-atime
14. Experimental Design: Choices
Play modified Ultimatum Game
Split own share of endowments between:
Household Account
Own Account
Spouse’s Account
Player A:
played 7- lotteries with 50% chance to win Rs. 75
Player B: no lottery option
Payoffs:
1.5
𝜋𝑖 = 𝑥𝑖 + 𝑔 ∗
2
16. Experimental Design: Player A
Flip Coin
Heads:
Extra Rs. 75
Private
&/or
Conceal
Split Endow
A+75, B, PG
Tails
Informed of Endowment
Distribution
Complete
&/or
Reveal
Split Endow+75
A, B, PG
Informed of Player B’s
Proposed Split
Accept or Reject
Split Endow
A, B, PG
17. Experimental Design: Player B
Private &
Conceal:
50% change
Player A has 75
Informed of
Endowment Distribution &
Outcome of Coin Toss
Split Endowment
A, B, PG
Informed of Player A’s
Proposed Split
Accept
Reject
Complete &
Reveal:
Player A has 75
If heads
18. Bargaining Power Treatment
Table 2: Distribution of Resources
Endowment
(Player B –
Player A) 1/
Distribution of Resources 2/
With Lottery Prize,
Without Lottery Prize
If known or revealed
Player B
255 - 45
85
150 - 75
67
195 - 105
65
150 -150
50
120 - 180
40
105 - 195
35
150 - 225
40
1/ Amounts in Indian Rupees.
Player A
Player B
15
33
35
50
60
65
60
68
50
52
40
32
28
33
Player
A
32
50
48
60
68
72
67
Change
(%)
17
17
13
10
8
7
7
Household
Endowment
300
225
300
300
300
300
375
19. Bargaining Power Treatment
Table 2: Distribution of Resources
Endowment
(Player B –
Player A) 1/
Distribution of Resources 2/
With Lottery Prize,
Without Lottery Prize
If known or revealed
Player B
255 - 45
85
150 - 75
67
195 - 105
65
150 -150
50
120 - 180
40
105 - 195
35
150 - 225
40
1/ Amounts in Indian Rupees.
Player A
Player B
15
33
35
50
60
65
60
68
50
52
40
32
28
33
Player
A
32
50
48
60
68
72
67
Change
(%)
17
17
13
10
8
7
7
Household
Endowment
300
225
300
300
300
300
375
20. Bargaining Power Treatment
Table 2: Distribution of Resources
Endowment
(Player B –
Player A) 1/
Distribution of Resources 2/
With Lottery Prize,
Without Lottery Prize
If known or revealed
Player B
255 - 45
85
150 - 75
67
195 - 105
65
150 -150
50
120 - 180
40
105 - 195
35
150 - 225
40
1/ Amounts in Indian Rupees.
Player A
Player B
15
33
35
50
60
65
60
68
50
52
40
32
28
33
Player
A
32
50
48
60
68
72
67
Change
(%)
17
17
13
10
8
7
7
Household
Endowment
300
225
300
300
300
300
375
21. Bargaining Power Treatment
Table 2: Distribution of Resources
Endowment
(Player B –
Player A) 1/
Distribution of Resources 2/
With Lottery Prize,
Without Lottery Prize
If known or revealed
Player B
255 - 45
85
150 - 75
67
195 - 105
65
150 -150
50
120 - 180
40
105 - 195
35
150 - 225
40
1/ Amounts in Indian Rupees.
Player A
Player B
15
33
35
50
60
65
60
68
50
52
40
32
28
33
Player
A
32
50
48
60
68
72
67
Change
(%)
17
17
13
10
8
7
7
Household
Endowment
300
225
300
300
300
300
375
25. Empirical Strategy: Baseline
𝑛
𝑔
𝑔
𝑔
𝑔
𝑥 𝑠,ℎ,𝑟 = 𝛿1 𝑇ℎ𝑃𝑟 + 𝛿2 𝑇ℎ𝑃𝑟−𝐷 + 𝛿3 𝑇ℎ𝑃𝑟−𝐶 + 𝛽
𝑤 𝑠,ℎ,𝑟 + 𝜃
𝑋 𝑠,ℎ,𝑟 +
7
𝛼𝑠 +
𝑠=1
𝜎 𝑟 + 𝜀 𝑠,ℎ,𝑟
𝑟=1
Dependent: Amount in HH Account or Own Account
Controls 𝑋 𝑠,ℎ,𝑟 : Gender, District and Lottery outcome (heads).
Own share of endowments 𝑤 𝑠,ℎ,𝑟
Round fixed effects.
Spouse random effects.
Hypothesis 1: 𝛿1𝑔 = 𝛿3𝑔 ≠ 0
𝑔
𝛿2 = 0
No guilty compensation
No differences among disclosers and PI
26. Results: Effect of Asymmetric
Information
Panel (a): Information Treatment Effects
HH Account
Own Account
All
Player A Player B
All
Player A Player B
Own Share
1.464*** 1.575*** 1.362*** 0.764*** 0.953*** 0.646***
Endowment
(0.043) (0.061) (0.061)
(0.036) (0.048) (0.047)
-7.628** -19.73*** 4.592
11.84*** 26.74*** -3.059
Private
(3.101) (4.221) (4.202)
(3.065) (3.308) (2.687)
2.312
3.639
0.920
-0.142
-0.613
-0.036
Private - Reveal
(3.469) (4.746) (4.326)
(2.464) (3.512) (2.775)
-10.46** -21.46*** 1.278
11.65** 31.88*** -7.040**
Private - Conceal
(3.642) (5.318) (5.132)
(5.214) (5.339) (2.568)
N
2799
1399
1400
2799
1399
1400
R-squared
0.4197
0.4449 0.4292
0.3078
0.4979 0.2446
Random Effects.
No differences across genders
27. Results: Effect of Asymmetric
Information
Panel (a): Information Treatment Effects
HH Account
Own Account
All
Player A Player B
All
Player A Player B
Own Share
1.464*** 1.575*** 1.362*** 0.764*** 0.953*** 0.646***
Endowment
(0.043) (0.061) (0.061)
(0.036) (0.048) (0.047)
-7.628** -19.73*** 4.592
11.84*** 26.74*** -3.059
Private
(3.101) (4.221) (4.202)
(3.065) (3.308) (2.687)
2.312
3.639
0.920
-0.142
-0.613
-0.036
Private - Reveal
(3.469) (4.746) (4.326)
(2.464) (3.512) (2.775)
-10.46** -21.46*** 1.278
11.65** 31.88*** -7.040**
Private - Conceal
(3.642) (5.318) (5.132)
(5.214) (5.339) (2.568)
N
2799
1399
1400
2799
1399
1400
R-squared
0.4197
0.4449 0.4292
0.3078
0.4979 0.2446
Random Effects.
No differences across genders
28. Baseline Results Description
As bargaining power increases:
Private Information:
Allocation towards HH account increases
Allocation towards Own account increases half as much
Decreases allocation towards HH account (Player A = information
advantage)
Statistically equal for those who choose to conceal
By construction allocation towards own account:
Increases for Player A
Decreases for Player B – knowing there is asymmetric information
increases sharing (other’s account)
29. Empirical Strategy: Two-way Interaction
3
𝑔
𝑥 𝑠,ℎ,𝑟 = 𝛽
𝑛
𝑔
𝑤 𝑠,ℎ,𝑟 +
𝑗
𝛿𝑗 𝑇ℎ × 𝑤 𝑠,ℎ,𝑟 + 𝜃
𝑗 =1
𝑋 𝑠,ℎ,𝑟 +
7
𝛼𝑠 +
𝑠=1
𝜎 𝑟 + 𝜀 𝑠,ℎ,𝑟
𝑟=1
Dependent: Amount in HH Account or Own Account
Controls 𝑋 𝑠,ℎ,𝑟 : Gender, District and Lottery outcome (heads).
Own share of endowments 𝑤 𝑠,ℎ,𝑟
Round fixed effects.
Spouse fixed effects.
Hypothesis 2:
𝑔
𝑔
𝑔
𝛿1 = 𝛿2 = 𝛿3 = 0
No differences in effect of
BP by Info. Treatment
30. Results: Two-way Interaction, BP
and Information
Panel (a):Two-way Interaction, (BP) and Information
HH Account
Own Account
All
Player A Player B
All
Player A Player B
Own Share
1.503*** 1.680*** 1.287*** 0.865*** 0.913*** 0.688***
Endowment
(0.084) (0.124) (0.106)
(0.060) (0.090) (0.061)
-0.001
-0.107
0.154
-0.170* -0.000
0.000
Private X Own Share
(0.110) (0.157) (0.150)
(0.089) (0.123) (0.100)
Private-Reveal X
-0.035
-0.110
0.075
-0.060
-0.017
-0.064
Own Share
(0.112) (0.160) (0.151)
(0.087) (0.119) (0.107)
Private-Conceal X
-0.170** -0.411** 0.007
0.128 0.529*** 0.126
Own Share
(0.086) (0.132) (0.110)
(0.107) (0.130) (0.097)
N
2799
1399
1400
2799
1399
1400
R-squared
0.4199 0.4466 0.4301
0.3086 0.4978 0.2426
Fixed Effects.
31. Results: Two-way Interaction, BP
and Information
Panel (a):Two-way Interaction, (BP) and Information
HH Account
Own Account
All
Player A Player B
All
Player A Player B
Own Share
1.503*** 1.680*** 1.287*** 0.865*** 0.913*** 0.688***
Endowment
(0.084) (0.124) (0.106)
(0.060) (0.090) (0.061)
-0.001
-0.107
0.154
-0.170* -0.000
0.000
Private X Own Share
(0.110) (0.157) (0.150)
(0.089) (0.123) (0.100)
Private-Reveal X
-0.035
-0.110
0.075
-0.060
-0.017
-0.064
Own Share
(0.112) (0.160) (0.151)
(0.087) (0.119) (0.107)
Private-Conceal X
-0.170** -0.411** 0.007
0.128 0.529*** 0.126
Own Share
(0.086) (0.132) (0.110)
(0.107) (0.130) (0.097)
N
2799
1399
1400
2799
1399
1400
R-squared
0.4199 0.4466 0.4301
0.3086 0.4978 0.2426
Fixed Effects.
32. Two-way Interaction Results
As bargaining power increases, allocation towards HH Account:
As bargaining power increases, amount towards Own Account:
Increases if Complete Information
No differences b/w Complete Information and Private & Disclose or
Private
Decrease if Private & Conceal : those least cooperative, become less
so as BP increases
Increases regardless of information environment
Except for:
Those who choose to conceal – increase allocation towards own
account significantly more.
33. Results: Two-way Interaction,
Gender and Information
HH Account
All
Player A Player B
Own Share
0.494*** 0.510*** 0.470***
Endowment
(0.013)
(0.018)
(0.020)
-11.94** -24.63***
1.494
Private
(4.146)
(5.551)
(5.834)
-1.675
0.534
-4.930
Private - Reveal
(4.277)
(5.842)
(5.273)
Private -9.839* -18.59**
0.260
Conceal
(5.420)
(7.669)
(7.919)
Gender
-3.127
-0.969
-4.727
(=1 if Male)
(4.379)
(5.950)
(5.576)
8.530
9.841
6.087
Private X Male
(5.938)
(8.181)
(8.057)
Private-Reveal
6.997
5.354
10.68
X Male
(6.606)
(9.209)
(8.171)
Private-Conceal 1.869
-3.156
4.691
X Male
(7.221)
(11.12)
(9.806)
Random Effects.
Own Account
All
Player A Player B
0.265*** 0.303*** 0.228***
(0.011)
(0.014)
(0.015)
14.55** 31.30***
1.419
(4.706)
(4.812)
(3.872)
4.248
4.745
1.868
(3.629)
(4.943)
(3.771)
11.42
29.22***
7.396
(8.059)
(7.895)
(6.568)
1.225
-1.192
-0.001
(3.250)
(4.535)
(3.757)
-5.479
-9.257
-5.299
(5.979)
(6.373)
(5.338)
-9.240*
-11.39*
-4.354
(4.824)
(6.785)
(5.491)
1.807
7.029
-6.126
(10.42)
(9.528)
(7.962)
34. Results: Two-way Interaction,
Gender and Information
HH Account
All
Player A Player B
Own Share
0.494*** 0.510*** 0.470***
Endowment
(0.013)
(0.018)
(0.020)
-11.94** -24.63***
1.494
Private
(4.146)
(5.551)
(5.834)
-1.675
0.534
-4.930
Private - Reveal
(4.277)
(5.842)
(5.273)
Private -9.839* -18.59**
0.260
Conceal
(5.420)
(7.669)
(7.919)
Gender
-3.127
-0.969
-4.727
(=1 if Male)
(4.379)
(5.950)
(5.576)
8.530
9.841
6.087
Private X Male
(5.938)
(8.181)
(8.057)
Private-Reveal
6.997
5.354
10.68
X Male
(6.606)
(9.209)
(8.171)
Private-Conceal 1.869
-3.156
4.691
X Male
(7.221)
(11.12)
(9.806)
Random Effects.
Own Account
All
Player A Player B
0.265*** 0.303*** 0.228***
(0.011)
(0.014)
(0.015)
14.55** 31.30***
1.419
(4.706)
(4.812)
(3.872)
4.248
4.745
1.868
(3.629)
(4.943)
(3.771)
11.42
29.22***
7.396
(8.059)
(7.895)
(6.568)
1.225
-1.192
-0.001
(3.250)
(4.535)
(3.757)
-5.479
-9.257
-5.299
(5.979)
(6.373)
(5.338)
-9.240*
-11.39*
-4.354
(4.824)
(6.785)
(5.491)
1.807
7.029
-6.126
(10.42)
(9.528)
(7.962)
35. Results: Three-way Interaction,
BP, Information and Gender
Own Share
Endowment
Private X Own Share
Private-Reveal X
Own Share
Private-Conceal X
Own Share
Private X Own Share
X Male
Private-Reveal X
Own Share X Male
Private-Conceal X
Own Share X Male
N
R-squared
Panel (b): Three-way Interaction, BP, Information and Gender
HH Account
Own Account
All
Player A
Player B
All
Player A
Player B
1.503***
1.681***
1.287***
0.865***
0.913***
0.688***
(0.084)
(0.124)
(0.106)
(0.060)
(0.090)
(0.061)
-0.105
-0.260
0.098
-0.115
0.159
0.206
(0.132)
(0.185)
(0.188)
(0.116)
(0.153)
(0.138)
-0.173
-0.308*
-0.005
0.123
0.209
-0.118
(0.145)
(0.183)
(0.219)
(0.120)
(0.135)
(0.171)
-0.183
-0.377**
-0.033
0.078
0.480**
0.207
(0.130)
(0.178)
(0.182)
(0.164)
(0.186)
(0.168)
0.208
0.314
0.107
-0.109
-0.330
-0.399
(0.142)
(0.187)
(0.212)
(0.133)
(0.163)
(0.151)
0.272*
0.401**
0.158
-0.359**
-0.460***
0.093
(0.149)
(0.191)
(0.223)
(0.123)
(0.140)
(0.182)
0.051
-0.046
0.100
0.077
0.077
-0.156
(0.169)
(0.262)
(0.225)
(0.210)
(0.237)
(0.194)
2799
1399
1400
2799
1399
1400
0.4218
0.4499
0.4311
0.3121
0.5034
0.2502
36. Three-way Interaction Results
Two-way interactions average out gender differences which
move in opposite directions
For Females, as BP increases:
Increase in allocation towards HH Account
No differences between Complete and Private Information
Decrease in allocation towards HH Account if Private Info with
Option relative to Complete Information
For Males, as BP increases:
No differences between Complete, Private Information or Private &
Conceal
Increase in allocation towards HH Account if Private & Reveal – Men
are more cooperative
37. Results: Three-way Interaction,
BP, Information and Gender
Own Share
Endowment
Private X Own Share
Private-Reveal X
Own Share
Private-Conceal X
Own Share
Private X Own Share
X Male
Private-Reveal X
Own Share X Male
Private-Conceal X
Own Share X Male
N
R-squared
Panel (b): Three-way Interaction, BP, Information and Gender
HH Account
Own Account
All
Player A
Player B
All
Player A
Player B
1.503***
1.681***
1.287***
0.865***
0.913***
0.688***
(0.084)
(0.124)
(0.106)
(0.060)
(0.090)
(0.061)
-0.105
-0.260
0.098
-0.115
0.159
0.206
(0.132)
(0.185)
(0.188)
(0.116)
(0.153)
(0.138)
-0.173
-0.308*
-0.005
0.123
0.209
-0.118
(0.145)
(0.183)
(0.219)
(0.120)
(0.135)
(0.171)
-0.183
-0.377**
-0.033
0.078
0.480**
0.207
(0.130)
(0.178)
(0.182)
(0.164)
(0.186)
(0.168)
0.208
0.314
0.107
-0.109
-0.330
-0.399
(0.142)
(0.187)
(0.212)
(0.133)
(0.163)
(0.151)
0.272*
0.401**
0.158
-0.359**
-0.460***
0.093
(0.149)
(0.191)
(0.223)
(0.123)
(0.140)
(0.182)
0.051
-0.046
0.100
0.077
0.077
-0.156
(0.169)
(0.262)
(0.225)
(0.210)
(0.237)
(0.194)
2799
1399
1400
2799
1399
1400
0.4218
0.4499
0.4311
0.3121
0.5034
0.2502
38. Three-way Interaction Results
For Females, as BP increases:
Increase in allocation towards Own Account
No differences between Complete, Private Information or Private &
Reveal
Increase in allocation towards Own Account if Private & Conceal by
50% more.
For Males, as BP increases:
No differences between Complete, Private Information or Private &
Conceal
Decrease in allocation towards HH Account if Private & Reveal – the
increase in HH account comes from reduction on own resources.
39. Mechanisms (female response)
workMembers of the HH that work for Income
for Income
Husband and Wife
Husband Only
Wife Only
ife Husband and Other than Wife
Dehradun
19
53
1
16
Almora
10
72
1
12
respondent decides
Proportion of HHs where respondent decides
Home
Whether to work outside of Home
How many children to have
s Major Household Purchases
es Gifts to respondent's relatives
Sell land
oney
Savings from Household Money
Dehradun
17
4
6
16
2
7
Almora
8
1
5
1
0
0
40. Conclusions
Theoretical Findings:
There exists a threshold level change of bargaining power that
needs to be exceeded to induce revelation
High Initial BP: threshold strictly positive
Low Initial BP: no incentives to conceal money
Empirical Findings:
When given the chance, 21% of spouses choose to conceal
No compensation through increased HH Good allocation
No differences by gender on Own and Public Good contributions
Men who choose to reveal are more cooperative than women:
As BP increases, men contribute more towards PG, while
women contribute less