Behavioural Economics Flashcards
Preferences
strict preference relation
indifference
Rationality
completeness
Transitivity
Choice based Approach
Revield preferneces or something is revealed to be prefereard to.
Weak axiom of reviled preferences
if x* reviled preferred over y* then y* can’t be reviled strictly preferred over x*
Reference-Dependent Preferences
u(c|r) = m(c) + n(c|r)
consumption utility + gain-loss-utility
where n(c|r) = \mu(m(c)-m(r))
\mu has
Loss aversion
diminishing sensitivity
Empirical Evidence for Reference Dependency
Taxi drives
Decoy-Effect
Endowment effect
Endowment Effect
Willingness to pay < willingness to accept
WTP < WTA
Limited Attention
The consideration set does not change if the irrelvant alternative is taken away.
Empirical Evidence Limited Attention
Left digit bias in the car retailer market.
Bayesian updating
P(A|B) = P(B|A)*P(A)/P(B)
Confirmation bias
Over weighting information that confirms the prior belief
Choices under Risk: Lottery
contains payoffs with associated risks
Preferences over lotteries
Continuity
Independence Axiom
Expected Utility in discrete case
sum over
Expected Utility in continuous case
integral with the cdf as dF(x) or with the pdf x*f(x) dx
Risk Aversion
C > EV
C = certainty equivalent
EV = expected value
Certainty Equivalent
U(C) = EU(L)
Prospect Theory
Relaxes the assumptions of the Expected Utility Theory:
- No-linear probability function
- utility over changes in wealth levels not over finite wealth levels
V(L) = Sum w(pi)*v(yi)
w(): weighting function
v(yi) value function
Allais Paradox
Same difference in prices changes the decission.
p = (25,5,0)
L1 = (0, 1, 0) L1’ = (0.1, 0.89, 0.01) -> choose L1 over L1’
L2 = (0, 0.11, 0.89) L2’ = (0.1, 0, 0.9) -> choose L2’ over L1
That is a contradiction thus not rational and can’t be explained with rational expectations.
Simple prospect theory fails to explain this.
Cummulative Prospect Theory
w(.) is applied to cummulitive probabilities.
Overcomes the Allais paradox
Measurement of risk preferences
Simple elicitation
Complex elicitation
Gneezy and Potters (1997)
initial endowment to be spent by part on a risky but under rational expectations profitable investment. If they did not spent everything they are risk averse.
Eckel and Grossmann (2002)
list of gambles each with 50% chance for high/low payoff.
Assuming a certain utility function allows - using the switching point to estimate a range for the risk parameter.
Holt and Laury (2002)
List of paired gambles with same payoff but changing probabilities.
Again flipping point gives the rage fro the risk parameter.
Problem, how to deal with inconsistent behaviour i.e. more than one switch?
Holt and Laury (2002)
List of paired gambles with same payoff but changing probabilities.
Again flipping point gives the rage fro the risk parameter.
Problem, how to deal with inconsistent behaviour i.e. more than one switch?
Decissions under Uncertainty
S: finite set of states of nature A: finite set of actions f:AxS -> R is the outcome function regret f(a_i,s_j) - max_s f(a_i,s) Maxmin criterion MaxMax criterion MinMax regrett criterion
Ellsberg Paradox
Urn with 90 balls 30 red 60 blue and green
Gamble 1:
1.L 100 if red
2.L 100 if blue
People tent to choose 1
Gamble 2:
- L 100 if ball is red or green
- L 100 if ball is blue or green
Most choose 2
This is inconsistent again. By choosing 1L in G1 they assumed implicit, that there are less then 30 blue balls., thus more then 30 green. But then red + green is more then 60 and blue and green is less then 60.
also called ambigiuity aversion
ambigiuity aversion
People don’t like gambles an prefer a smaller certain outcome.
Halevy (2007)
Four lottereis of which two are just compound lotteries that gice the same lottery as the first. Thus all three should be evaluted the same.
Results:
People showed ambigiuity aversion and not all valued the three same lotteries the same way.
Exponential Discounting
time consistent discounting process but does not explain emperical findings
Quasi-Hyperbolic discounting
Like exponential with an additional constant discount factor from t >1 onwards.
Loewenstein and Kalyanaraman (1999) Movie Rental
When people where to choose movies in advance vs. on that day.
More likely to choose “high brow” movies when choosing in advance.
Van Leenwen (1998) Snack Choices
People had to choose snacks. There were different treatments. Choosing when hungry or satisfied and in advance or at that time as well as all combinations.
Results:
healthier chices in advance when people where satisfied
Niederte and Sprenger (2015) Allocation Effort
Effort fro simple tasks
- Allocation is made on the first day
- Allocation sunsequently on each day.
Result in the subsequent allocation treatment the individuals got less work done.
=> make a plan and stick to it.
Measuring time preferences - Double Multiple Price listing
price pairs one small-soon and high-late payments
the switching point from small-soon to high-late gives the interval for the estimation of the time preference parameters
Time independent constant rat of risk aversion
CRRA
Convex Time Budget (CTB)
Allowing subjects to choose any payoff between the tow boundaries of the experiment.
Degree of Self Awareness
What is the discount factor in the future, compared to today.
ß = ^ß = 1 -> exponential discounting
ß < ^ß = 1 -> Naive quasi hyperbolic discounting
ß = ^ß < 1 -> Sophisticated quasi hyperbolic discounting
O’Donoghue and Robin (1999) Procrastination
percetion-perfect strategies solved with backward induction and gives the optimal strategy.
Even when the path ends early the strategy includes actions for all possible periods.
Gul and Pesendorfer (2001) Temptation and Self Controle
Two choices:
- Menu choice in advance which restaurant to go to
- Choice from the menus available after 1.
three cases:
- Case no tmptation
- Case temptation
- Case self-controle
Beck, De Groot and Marschak (1964) WTP elicitation
asking for the valuation and then choosing the price randomly.
causes an incentive compatible result
Individuals have an incentive to report their true valueation
Elicitating probabelistic beliefs
how to get the true probabilities people base their actions on?
Scoring rule: Quadratic scoring rule (works only under risk neutrality)
Scoring Rule
Relates the reported probability to a given payoff
Hossain and Okni (2013) Porb. beliefs elicitation
Binanized Scoring Role
if A:
1-(1-r)^2
(1-r)^2
if not A:
1-r^2
r^2
Libertarian Paternalism
soft behavioural interventions that are not interfearing with the individuals liberty but use psychological effects to “force” them to do the right thing.
Choi, Laibson and Madrian (2004) Retirement savings
Problem: Present bias leads people to neglect their pension payments. As well as a reference dependency and a status quo bias.
Solution:
automatic enrollment or offer when hired.
both show a significant higher number of participants but the automatic enrollment has the highest.
Cooling off periods
Fully ration undertake actions only if y>0
but if we have bounded rationality y + e > 0 where e is a random error
with a cooling off period we reduce the value of the good y* < y but we could:
Potential gain: y+e > 0 > y > y*
Potential costs:
- y > y* > 0 - action taken anyway just slightly lower value
- y > 0 > y* will not act due to cooling off but value of good must have been quite small or cool down effect is very strong.
