Week 8
Incentives, Motivation and Changing Behaviour
Gneezy and Rustichini (2000)
Method
• 10 private day care centre in Israel that shut at 4pm
• Study took place over 20 weeks
◦ Weeks 1-4: monitor number of late collections per week
◦ Week 5: ne for lateness (per child being collected) introduced in 6 of the centres, selected
randomly
◦ Week 17: ne removed without explanation
• Traditional hypothesis is that a ne would reduce this behaviour
Results
• In the test group the number of occurrences increased
steadily in weeks 3-4 to a level almost twice as high
as the initial one
• After the ne was removed the no. late arrivals
remained stable
Game theory explanation
This is a model of di erential information and incomplete contracts
• Without ne, parents learn that the penalty for occasional lateness is not severe and an implicit
contract is formed
• The introduction of the ne although none of the explicit terms of the contract have changed it
provides more information and believe the ne is the worst that can happen, so the number of
delays increases
• Once the ne is removed they have no reason to anticipate anything more than a ne so do not
reduce latenesses
The model:
• Two types of nursery owner
◦ Type S, who can kick a child out of the day care, action K
◦ Type M, who can raise a ne, f
◦ Both types can also do nothing
◦ Type is unknown to the parents, and p(S)=p(M)=0.5
• Sequential game
◦ The owner chooses the action, if they play f or K they have to pay a xed implementation
cost for the rst period
◦ Parents simultaneously choose an amount of delay
◦ If K is played then the game is over for that parent
• Payo s, all discounted
◦ Parents: delay length x (value of delay, v, - f), where v>f
, ◦ Owner, M : delay length x (f - cost of delay)
◦ Owner, S: 0
‣ And the cost of playing f or K discounted to the period it was implemented
Equilibrium:
• Parents
◦ Parents choose an around of delay, d* at which the owner f,K~ nothing
‣ d>d* is deferred by the expected loss if K occurs
◦ If a parent does play d>d* and the ne is introduced then all players start maximising delay
time, as if they are going to be late at all they will be ned so might as well maximise their
delay time as v>f
• Owner
◦ Does nothing in each period as long as d* is not exceeded in previous periods and they have
been doing nothing
◦ If any player, including themselves, has deviated in previous period she plays f,K
‣ M types know that after revealing their type parents will start choosing a larger delay so
might as well maximise compensation by introducing the ne
‣ S types know that if they do nothing parents will think they are M type and increase their
delay, so will cut their losses and choose K
This yields the predictions that:
• In the initial period there is a low level of late arrivals
• If o the equilibrium path the owner charges a ne, then parents know they are type M and
increase the number of delays
• If the ne is removed the parents still know this so do not change behaviour
A larger and proportional fee (rather than at fee) could overcome this outcome
Social norm explanation
• The introduction of the ne changes perceptions
◦ The parents considered lateness without a ne a kindness, so don’t wish to take advantage
of it
◦ Once the ne is introduced it is viewed as a service to be bought
• This is due to a social norm of accepting uncompensated help with restraint, but help at price can
be used as wanted
• Once the ne is removed the help is still viewed as a service just at a price of zero, so is still
bought in large quantities
Falk and Kosfeld (2006)
SETUP
Vance
N PA Agent
Principal
Up x
s t.cz trustcontrolasomFeithermonitoring
sits orcontractual
810
Incentives, Motivation and Changing Behaviour
Gneezy and Rustichini (2000)
Method
• 10 private day care centre in Israel that shut at 4pm
• Study took place over 20 weeks
◦ Weeks 1-4: monitor number of late collections per week
◦ Week 5: ne for lateness (per child being collected) introduced in 6 of the centres, selected
randomly
◦ Week 17: ne removed without explanation
• Traditional hypothesis is that a ne would reduce this behaviour
Results
• In the test group the number of occurrences increased
steadily in weeks 3-4 to a level almost twice as high
as the initial one
• After the ne was removed the no. late arrivals
remained stable
Game theory explanation
This is a model of di erential information and incomplete contracts
• Without ne, parents learn that the penalty for occasional lateness is not severe and an implicit
contract is formed
• The introduction of the ne although none of the explicit terms of the contract have changed it
provides more information and believe the ne is the worst that can happen, so the number of
delays increases
• Once the ne is removed they have no reason to anticipate anything more than a ne so do not
reduce latenesses
The model:
• Two types of nursery owner
◦ Type S, who can kick a child out of the day care, action K
◦ Type M, who can raise a ne, f
◦ Both types can also do nothing
◦ Type is unknown to the parents, and p(S)=p(M)=0.5
• Sequential game
◦ The owner chooses the action, if they play f or K they have to pay a xed implementation
cost for the rst period
◦ Parents simultaneously choose an amount of delay
◦ If K is played then the game is over for that parent
• Payo s, all discounted
◦ Parents: delay length x (value of delay, v, - f), where v>f
, ◦ Owner, M : delay length x (f - cost of delay)
◦ Owner, S: 0
‣ And the cost of playing f or K discounted to the period it was implemented
Equilibrium:
• Parents
◦ Parents choose an around of delay, d* at which the owner f,K~ nothing
‣ d>d* is deferred by the expected loss if K occurs
◦ If a parent does play d>d* and the ne is introduced then all players start maximising delay
time, as if they are going to be late at all they will be ned so might as well maximise their
delay time as v>f
• Owner
◦ Does nothing in each period as long as d* is not exceeded in previous periods and they have
been doing nothing
◦ If any player, including themselves, has deviated in previous period she plays f,K
‣ M types know that after revealing their type parents will start choosing a larger delay so
might as well maximise compensation by introducing the ne
‣ S types know that if they do nothing parents will think they are M type and increase their
delay, so will cut their losses and choose K
This yields the predictions that:
• In the initial period there is a low level of late arrivals
• If o the equilibrium path the owner charges a ne, then parents know they are type M and
increase the number of delays
• If the ne is removed the parents still know this so do not change behaviour
A larger and proportional fee (rather than at fee) could overcome this outcome
Social norm explanation
• The introduction of the ne changes perceptions
◦ The parents considered lateness without a ne a kindness, so don’t wish to take advantage
of it
◦ Once the ne is introduced it is viewed as a service to be bought
• This is due to a social norm of accepting uncompensated help with restraint, but help at price can
be used as wanted
• Once the ne is removed the help is still viewed as a service just at a price of zero, so is still
bought in large quantities
Falk and Kosfeld (2006)
SETUP
Vance
N PA Agent
Principal
Up x
s t.cz trustcontrolasomFeithermonitoring
sits orcontractual
810
