Generosity and social distance in dictator game field experiments with

and without a face By

Sosina Bezu and Stein Holden

School of Economics and Business

Norwegian University of Life Sciences

www.steinholden.com

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Introduction The dictator game may be the simplest tool for

investigating generosity (Dufwenberg and Muren 2006)

There are mixed evidences on the gender differences and

effects of social distance on generosity (Eckel and

Grossman 1998; Dufwenberg and Muren 2006)

– Could be due to variations in methods and framing

effects

Do findings in lab experiments in the West generalize to

poor people in developing countries?

– Contexts where women have a much weaker social

position

– Family networks may be more important in such contexts

and affect the extent of generosity and interdependence

of preferences

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Theoretical model

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

,i i

j j ig g

i

v X xU U x

d

Building on Fehr and Schmidt (2005): Taking into

account social preferences and interdependent

preferences:

Individual and interpersonal characteristics of the

person ( )

Marginal utility effect on the other party,

Social distance (d )

Gender-specific preferences (g )

j

j iv X x

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Hypotheses (1) a) Family members are more likely to receive positive

amounts/receive larger amounts in dictator games than

anonymous villagers from the same village (effect of social

distance)

b) Generous family members (spouses) are more likely to get

positive amounts (and receive larger amounts) than less

generous family members (spouses) (interdependent

preferences)

c) Persons who are more generous towards anonymous

villagers are also likely to be more generous towards their own

family members and vice versa (social preferences)

d) Exposure to generous family members contributes to

generous behavior towards anonymous persons (responsive

social preferences)

4

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Hypotheses (2)

e) Women are more generous to their husbands and

anonymous villagers (have stronger social preferences)

than men are.

f) Spouses that themselves selected each other through

voluntary marriage contracts (love marriage) are more

generous towards each other in dictator games than

spouses that were married through involuntary or

enforced marriages

5

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Sampling and data

A sample of 600 households in 17 villages in Southern

Ethiopia surveyed in 2007 and 2012 to investigate

impacts of a land tenure reform aiming to strengthen

household tenure security and women’s land rights

through joint land certification of husbands and wives

Social experiments: Dictator games in 2012

– With husbands and wives in monogamous households:

385 households

– With husbands and wives in polygamous households: 90

households

– With one other family member in remaining households

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

6

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Experimental design, 1

Experiment took 1 day in each village

Designed to prevent communication between households

and household members during the experiment

– Separate play room and two waiting rooms/places

– Household couples were called one at the time

– Coin toss (1) to identify a “winner” to play first, the loser

had to wait in the other waiting room till all “winners”

had played

– The winner was given 40 EB and two stated preference

questions for willingness to share with the waiting family

member/spouse and with an anonymous person in the

sample of households in the village

– Coin toss (2) to determine whom to share with of the two

7

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Experimental design, 2

First player receives her/his money and amount given to

family/anonymous person is put in an envelope

Player is asked to hide the money, go home an not talk to

anybody about the game or what was won

Same procedure for all pairs

Losers are called one by one in the same order

Are asked the same “what if” stated preference (hypothetical)

questions about what they would allocate to spouse/family

member and anonymous person

They are given the envelope that either comes from their

spouse or anonymous person with an amount topped up with

10 EB

They are asked to hide the envelope, and go home without

talking to anybody

DETTE E

R T

ITTELEN

PÅ P

RESEN

TASJO

NEN

8

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Example of location for field experiments

9

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Example of game environment

10

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Another “field lab” location

11

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

DETTE E

R T

ITTELEN

PÅ P

RESEN

TASJO

NEN

12

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Allocation to spouse by husbands and wives

020

40

60

0 10 20 30 40 0 10 20 30 40

Husbands Wives

Pe

rce

nt

How much money allocated, EB

Graphs by Gender of player

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Allocation to anonymous villager

020

40

60

80

0 10 20 30 0 10 20 30

Husbands Wives

Pe

rce

nt

How much money allocated, EB

Graphs by Gender of player

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Mean allocations by gender and receiver

15

Allocation to spouse Allocation to

anonymous villager

Probability of

non-zero

allocation

Average

amount

allocated

Probability

of non-zero

allocation

Average

amount

allocated

Husbands Mean 0.850 16.816 0.291 2.940

St. error 0.018 0.462 0.023 0.283

N 380 380 385 385

Wives Mean 0.766 14.408 0.255 2.455

St. error 0.022 0.448 0.022 0.251

N 380 380 385 385

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Validity of within-household experiments?

A unitary household model would

invalidate the experiments as households

would share the cash after the game

anyway

We assessed the extent of sharing of cash

through separate questions to the spouses

about their ability to mobilize cash for an

urgent need and the extent to which they

could obtain this cash from their spouse

The results follow:

16

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Ability to mobilize cash and extent of help from spouse

17

Wives’

cash

mobili-

zation

Get

cash

from

hus-

band

If cash

from

husband,

how

much

Husbands’

cash

mobili-

zation

Get

cash

from

wife

If cash

from

wife,

how

much

Mean, EB 491.64 0.20 145.93 1219.85 0.02 325.00

Median,

EB

200 0 100 500 0 150

St. error 118.00 0.02 17.49 185.60 0.01 150.42

N 366 367 75 352 359 6

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

Indicates clearly limited pooling of cash among

spouses

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Generosity and interdependent preferences, linear probability models with village fixed effects

18

Wife

gives to

husband,

dummy

Husband

gives to

wife,

dummy

Wife gives to

anonymous,

dummy

Husband

gives to

anonymous,

dummy

Real game dummy 0.031 0.076** -0.024 -0.042

(0.040) (0.030) (0.040) (0.050)

Received positive amount 0.312**** 0.018

from husband, dummy (0.060) (0.060)

Received positive amount 0.207**** 0.083

from wife, dummy (0.040) (0.060)

Giver to anonymous, 0.219**** 0.162****

dummy (0.050) (0.040)

Giver to spouse, dummy 0.257**** 0.311****

(0.050) (0.070)

Constant 0.425**** 0.609**** 0.058 -0.016

(0.060) (0.040) (0.070) (0.070)

Prob > chi2 0.000 0.000 0.000 0.000

R-squared 0.128 0.137 0.068 0.075

Number of observations 380 380 385 385

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Wives’ allocation to anonymous villager vs. zero or non-zero allocation to spouse

19

050

10

0

0 5 10 15 20 0 5 10 15 20

0 1

Perc

ent

How much money allocated by wife to anonymous

Graphs by dummy:player gave nonzero amount to spouse

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Husbands’ allocation to anonymous villager by zero or non-zero allocation to spouse

20

050

10

0

0 10 20 30 0 10 20 30

0 1

Perc

ent

How much money allocated by husband to anonymous

Graphs by dummy:player gave nonzero amount to spouse

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Allocation by wives to husbands by zero or non-zero amount given to anonymous villager

21

020

40

60

80

0 10 20 30 0 10 20 30

0 1

Perc

ent

How much money allocated by wives to husbandGraphs by dummy: player gave nonzero amount to anonymous

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Allocation by husbands to wives by zero or non-zero amount given to anonymous villager

22

020

40

60

80

0 10 20 30 40 0 10 20 30 40

0 1

Perc

ent

How much money allocated by husband to wife

Graphs by dummy: player gave nonzero amount to anonymous

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Allocation by wives to husbands by zero or non-zero amounts received from husbands

23

020

40

60

80

0 10 20 30 0 10 20 30

0 1

Perc

ent

How much money allocated by wife to husband

Graphs by Dummy:Non-zero amount received from husband

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Allocation by husbands to wives by zero or non-zero amount received from wife

24

020

40

60

80

0 10 20 30 40 0 10 20 30 40

0 1

Perc

ent

How much money allocated by hysband to wife

Graphs by Dummy:Non-zero amount received from wife

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Correspondence of stated marriage type by spouses

25

Marriage type: females’

responses

Marriage type:

males’ responses Parental

arrange

ment

Parental

and bride

agree-

ment

Love

marri-

age

Wife

ex-

change

Kidnap-

ping

marri-

age

Total

Parental arrangement

60 7 11 1 4 83

Parental and bride

agreement 14 81 7 1 3 106

Love marriage 11 12 132 0 11 166

Wife exchange 0 1 2 5 0 8

Kidnapping marriage 1 1 2 0 18 22

Total 86 102 154 7 36 385

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Models with wider sample, incl. polygamous hhs and other hh members than the spouse

26

Prob. of non-zero allocation Non-zero amounts

allocated

Linear probability models Truncated tobit models

Wives Husbands Wives Husbands

Dummy for real game, -0.018 0.023 -1.541*** -0.573

Allocation for whom: Baseline=Allocation for spouse

Son or daughter 0.115 -0.009 -2.478 0.888

Mother or father 0.153 . -1.104 .

Other relative 0.322**** 0.432 -2.442 0.211

Wife 2-4 (polygamous hhs) -0.215**** -0.020 -7.125**** -2.521****

Anonymous person -0.492**** -0.558**** -9.251**** -9.290****

Type of marriage: Baseline=Parental arrangement

Parental and bride agreement -0.053 -0.013 -1.678** -1.904**

Love marriage -0.041 0.020 -1.522** -1.449*

Wife exchange -0.003 -0.105 -0.337 1.883

Kidnapping marriage -0.083 -0.148* -3.911**** -1.540

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Summing up: Hypothesis testing/Conclusions 1

Hypothesis a) that family members are more likely to receive

positive amounts/receive larger amounts in dictator games than

anonymous villagers from the same village (effect of social

distance); cannot be rejected;

Hypothesis b) that generous family members (spouses) are

more likely to get positive amounts than less generous family

members (spouses) (interdependent preferences), cannot be

rejected;

Hypothesis c) that persons who are more generous towards

anonymous villagers are also likely to be more generous

towards their own family members and vice versa (social

preferences); cannot be rejected;

Hypothesis d) that exposure to generous family members

contributes to generous behavior towards anonymous persons

(responsive social preferences), was rejected;

27

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face

NO

RW

EG

IAN

UN

IVERSIT

Y O

F L

IFE S

CIE

NCES

www.umb.no

Summing up: Hypothesis testing/Conclusions 2

Hypothesis e) that women are more generous to their

husbands and anonymous villagers (have stronger social

preferences) than men are; was rejected;

Hypothesis f) that spouses that themselves selected each other

through voluntary marriage contracts (love marriage) are more

generous towards each other in dictator games than spouses

that were married through involuntary or enforced marriages;

was partly rejected.

Wives married through love marriages and through parental

and bride agreement were less generous than wives married

through parental arranged marriages.

Wives married through kidnapping marriages were less

generous towards their husbands, in line with the hypothesis.

28

Genero

sity a

nd so

cial d

istance

in d

ictato

r gam

e fie

ld

experim

ents w

ith a

nd w

ithout a

face