Economic Theories of the Household – Some Experimental

Love And Money: Experimental
Tests Of Theories Of The
Household.
Ian Bateman,* Tara McNally** and Alistair Munro**
*University of East Anglia, Norwich, UK
** Royal Holloway College, University of London.
http://personal.rhul.ac.uk/UQTE/003/Couples.htm
What are we testing?
• Theories of the household, notably:
– Income pooling – does the source of income affect
choices?
– Pareto efficiency
– Unitary models
– (theories of risky choice)
– Altruistic behaviour
What do we do?
• We
– Recruit established couples
– Separate them
– Make them do choices separately
• Most of these choices involve gambles…
• E.g. £20 for me versus a 60% chance of £40 for my
partner
– We make them predict their partner’s choices
– Re-unite couples and face them with joint choices.
– Pick one of their 36 choices at random and make them
play it out.
How do we run the experiment?
Subjects enter, allocated
‘wave’ or ‘triangle role.
Sub-groups separated
Triangles: Individual
choice. Questions 1...n
Waves: Individual choice
Questions n+1...2n
Section 2: Prediction,
Questions n+1...2n
Section 2: Prediction,
Questions 1...n
Groups merge, Joint choice
Questions 2n+1...3n
One question chosen at
random, choice executed, payoffs
Example task & test
Does information matter?
•
•
•
•
•
Treatment 1 (n=76),
separated couples in
different rooms,
no revelation of
answers to partner;
private and separate
payouts
in sealed envelopes.
These facts preannounced
•
•
•
•
Treatment 2 (n=40)
Separated couple in
same room
All answers available to
partner
Payments made in front
of partner
These facts preannounced
Within and between task tests for
consistency with Income Pooling
100%
90%
80%
70%
60%
50%
Separate
Within (Informed)
Between (Informed)
Joint
Within 2 (Informed)
Between (Private)
Within 2 (Private)
Income Pooling. Typically…
• cannot reject income pooling for joint choice
– Pattern of deviations consistent with inequality
aversion
• For separated choice
– Reject income pooling for separated choice
– Pattern of deviations consistent with greater
weight placed on own payoffs.
• Information regime has little impact.
Pareto efficiency
30
Number where
• Across a wide range of
tasks couples who
make the same choice
separately, deviate from
that choice in joint
deliberation
• Deviation is systematic
– away from risky
choices.
40
20
10
0
1
2
4
7
9
Task
Individuals differ
All agree
Couples differ
11
13
Summing up
• Strong rejection of unanimous decisions
• Rejection of income pooling in separated choice
• Widespread deviation from Pareto efficiency
– Couples are more risk averse than the individual partners
• Results are robust to information disclosure rule
– Though some evidence that women are less risk averse with
information disclosure
• Suggestion is that standard household models are
inappropriate.
• More details on this sequence of experiments at:
• http://personal.rhul.ac.uk/UQTE/003/Couples.htm