8 - explanations of Allais paradoxes Flashcards
what is the other explanation of CCE and CRE than prospect theory
salience theory
what is salience theory
- besides from probabilities - decision weights attached to events depend on difference in outcomes of the available option in relevant events
- options are evaluated by comparing to other options
- decision = which option is better in the event where options differ the most
what is salience theory distinct prediction
CCE will happen if the lotteries are stochastically independent and wont happen if they arent
the larger the difference between the options in relevant event…
the larger the decision weight on the event will tend to be
- the more attention is given to that option
what will we do
for simplicity we approximate this by supposing choice between options by comparing them in the event where they differ the most
what are the 2 interpretations of this
- attention based
- regret based
what is attention based
- decision maker has limited attention
- draws its attention to the event where its decision will matter the most - where there is a biggest difference
what is regret based
- decision maker is averse to large regrets so acts to avoid the possibility of them
- doesnt want to regret what might have been by not choosing the better option in the event that differs the most
- biggest scope for regret in the most different event
what is a stochastically independent option
each option is resolved independently
- will roll the die once for option 1 and another time for option 2
- the outcomes are uncorrelated
what is a stochastically dependent option
one roll resolves both options
* outcomes are correlated
how can we apply salience theory to Allais paradoxes
what changes need to be made
- lotteries are resolved by draws from urn with balls numbered 1-100
- replace probabilities with events defined by draws
what is the CCE when there is a single draw that resolves both options
stochastically dependent = one ball resolves both of the problems
- rules out the CCE - because safest options are chosen in both problems
- because looks at event with the biggest difference in outcomes and chooses best option = safe option has 2400, risky has 0
what is CCE when 2 draws
stochastically independent
options are resolved by different balls
- 1 resolves safe, 1 resolves unsafe
CCE is restored
- choose safe in first option and risky in second option
when wont CCE happen
when will it happen
when options in each choice are stochastically dependent
when stochastically independent
what does salience theory predict
that CCE can occur if options are resolves separately - stochastically independent
and wont occur when options are stochastically dependent
why doesnt the way that lotteries are resolved matter for EUT or prospect theory
prospect theory = decision weighting is driven by probabilities within each lottery
salience = decision weighting driven by comparison across lotteries - matters how it is resolved
when will you choose the risky in problem 2
if more emphasis is placed on the event with the biggest difference - this dominates decision making - and you pick the riskier option
what did the Bruhin et al 2022 study exploit
exploits how lotteries are resolved to discriminate between prospect theory and salience theory
- prospect = resolution doesnt matter
- salience = only see CCE in independent
what was the Bruhin experimental design
- subjects face 2 blocks of choices
- 1 block = lotteries that are stochastically dependent
- 1 block = lotteries that are stochastically independent
- order faced randomised
what are the Bruhin predictions
EUT
* subjects will not display CCE or CRE
prospect theory
* subjects may face CCE and CRE, but will be equally prone regardless of whether lotteries dependent/independent
salience theory
* may display CCE and CRE but only when lotteries are stochastically indepednet
Bruhins
how were they displayed
- proportion of observations where violations of EUT observed - presents the expected direction (CCE/CRE) and opposite direction
- also presents the violations of independent and dependent results separately
what were Bruhins results
evidence of violations of EUT
but what direction
- 30% violations in expected direction
- also violations in unexpected direction
- lots of people that are EUT consistent
Bruhins results for salience theory
- more violations of EUT in expected direction when payoffs stochastically independent = as predicted
- but still cases of CCE/CRE in dependent = what prospect theory allows for - salience doesnt account for this
what would you expect results to be for salience theory
- expected direction of EUT violations only present in indepedent and 0 for dependent
- but this is not the case
- even though indepednent > dependent
Bruhin results that evaluate CCE and CRE separately
- bigger difference in independent and dependent frequencies for CCE than CRE
Bruhin conclusions
i guess evidence for both salience and prospect?
cons of Bruhin experimental design
within-subjects feature
all subjects face all of the tasks
- boredom
- perception each choice will not be the paid one