lecture 1
1.2 Assumptions of standard economics
● Structure of preferences over outcomes (axioms like transitivity, completeness etc.)
● Only outcomes matter (final wealth) not changes, not intentions
● Belief updating and subjective probability (doesn’t matter what the intention is)
● No (or negligible) cost for information or decision costs (effort and time)
● No limits of reasoning power
● No self control problems
1.3 Examples of violations of standard assumptions
1. consumer preferences
- Completeness: if we have two bundles A and B then either A is preferred to B, or B is
preferred to A, or the consumer is indifferent between A and B. “I don’t know” is not
allowed!
- Transitivity: if A is preferred to B and B is preferred to C than A is preferred to C.
Completeness: you know what you like
- If you have trouble to choose, you are apparently indifferent
- But that means that if a very little extra is added to one of the options, that option will
be preferred.
- If one of the suitors would tell her: “if you choose me I will buy you a bicycle”: would
that help?
- If individuals have trouble choosing,
they tend to:
● Postpone a decision if possible
● Stick to the current situation if possible
● Accept the default if available
Status Quo bias: stick to the current situation
Default bias: the tendency to make the choice for which no action has to be taken
Transitivity
A scores better than B on Time and Teacher
B scores better than C on Time and Content
But... C scores better than A on Teacher and Content
violation of transitivity →
Condorcet paradox In many cases the decision maker is not an individual but a group (e.g.
family, firm). In these cases Condorcet cycles (you stay in a cycle) can exist.
A committee with three members have to vote on three options; first two options and than
the winner against the third one. The outcome depends on the order of votes
If member 1 (point in time) is chairman: B versus C, A against winner: A wins
If member 2 (teacher) is chairman: A versus B, C against winner: C wins
If member 3 (content) is chairman: A versus C, B against winner: B wins
,why intransitivity is bad →
Money machine to extract money from decision makers with intransitive
preferences:
If A is preferred to B, B is preferred to C and C is preferred to A;
- The consumer possesses A.
- Offer: if you pay 1 cent you can change your A for C
- The consumer pays 1 cent and now possesses C
- Offer: if you pay 1 cent you can change your C for B
- The consumer pays 1 cent and now possesses B
- Offer: if you pay 1 cent you can change your B for A
- The consumer pays 1 cent and now possesses A
Result: the consumer loss 3 cents and again possesses A; we can repeat this until the
consumer is bankrupt or makes his preferences transitive!
In many situations outcomes are uncertain
Risk: probabilities are known (throw a 6 with a standard dice)
Ambiguity: probabilities are unknown (e.g. the Dow Jones will be >40000 on December 31
of this year)
Ambiguity is much more common than risk in real life
Standard Econ with risk: Expected Utility
Two more axioms needed:
- Continuity: If I prefer 4 apples over 4 pears, I should also prefer 2 apples and 2
pears over 4 pears. (If X is preferred over Y then aX+(1-a)Y will be preferred over Y
for all 0<a<1)
- Independence of irrelevant alternatives: If X ∼ Y (the decision-maker is indifferent
between X and Y) then for all Z and p [p:X , (1-p):Z] ∼ [p:Y , (1-p):Z]
(the decision-maker is indifferent between a lottery with prize X with probability p and
a prize Z with probability (1-p) and the lottery with prize Y with probability p and a
prize Z with probability (1-p)). The idea is that in both lotteries the probability (1-p) to
receive Z is the same, and the lotteries only differ in the prize that is received with
probability p (respectively X and Y) so only the preference for X and Y should
influence the preference between the two lotteries.
If these 4 (Completeness,Transitivity, continuity and Independence of irrelevant alternatives)
hold → there exists an Expected Utility function such that:
U(p:X , (1-p):Y)=pU(X) + (1-p)U(Y)
In EU two psychological properties
- Diminishing/increasing sensitivity for money
- Risk attitude risk averse/risk loving are the
same!
, Standard Econ with ambiguity: Subjective Expected Utility
- When the probabilities are not known standard economics assumes that the
decisionmaker attaches a subjective probability to each possible event.
- Needed assumptions:
● Subjective probabilities add up to not more than 1
● If A is a subset of B, then P(A)≤P(B)
● The probabilities are independent of the consequences of the event (no optimism or
● pessimism allowed!)
Violation of this assumptions:
❖ Independence of irrelevant alternatives in EU
- What do you prefer:
A: 30 euros with probability 1
B: 40 euros with probability .80, 0 with probability .20
- What do you prefer:
C: 30 euros with probability .25, 0 with probability .75
D: 40 euros with probability .20, 0 with probability .80
- Most prefer A over B, but D over C. But:
E: play lottery A with probability .25, 0 with probability .75
F: play lottery B with probability .25, 0 with probability .75
E=C and F=D !
❖ Optimism
the subjective probability is larger for events that are positive for the decision-maker. In
Western culture (especially North-American) optimism is seen as a positive personality
trait, while to economists it is considered a bias (which can lead to suboptimal decisions).
❖ Additivity biases
a) How likely is it that the Dow Jones will end December 31 between 40000 and 42000?
b) How likely is it that the Dow Jones will end December 31 between 42000 and 45000?
c) How likely is it that the Dow Jones will end December 31 between 40000 and 45000?
Typically a+b>c
❖ Availability
- The subjective probability of an event is in practice based upon how easy it is to
imagine the event or to retrieve past instances
- Ajax plays against an amateur club. How likely is it that:
1. Ajax will get 1-0 behind
2. Ajax will get 1-0 behind, but in the end will win
- Pr(2) should be smaller than Pr(1), but Pr(2) could be estimated more
likely than Pr(1)
, 1.4 bounded rationality
- Standard theory assumes that information is costless, information processing is
costless, the agent optimizes and make no errors
- In reality people make errors (and they know that), decision costs can be large
compared with the cost of errors and many problems are to complicated. Therefore
agents simplify problems and look for a solution that is good enough.
Rational decision making “simple” case (no uncertainty)
- Alternatives: What are the possible actions I can take and to what outcomes do they
lead?
- criterions: What are the criterions to compare the outcomes?
- Relative importance of criterions: What is the relative importance of each criterion?
- Calculate: Analyze each action/outcome, using weighted criterions
- Choose: Choose the best action
the assumption that we know our preferences is not realistic, Do we have to find out all
relevant characteristics for all these houses? Too many options → simplify the problem:
filtering alternatives, formulating a minimum requirement. (fewer criterions, fewer
alternatives)
Herbert A. Simon Started with studying decision processes within firms:
1. Many problems are too complex, or not worthwhile
2. Firms searches for a solution that is “good enough”
3. Firms simplify the problem by formulating sub-goals
4. Use of heuristics and tradition (how did we do it in the past, and can we use that)
5. Change behavior only if you are not happy with the outcomes, search locally for
improvement. (Only when a problem is detected, the search for a solution will start)
6. Decision tasks are distributed, coordination is by communication and hierarchy
The first 5 points are also relevant for individual decision makers
Intuition and learning
How you simplify the problem and which action you consider depends on characteristics
of the situation.
- An expert is much better in this than a starter → But also an expert cannot tell you
how he does this
- Intuition is based upon experience
- Learning is easier with direct (time) and univocal (good or bad, easy to feedback
does not really change) feedback
- A problem is that learning by experience is asymmetric → The decision maker
chooses an action, gets feedback and improves his knowledge about the quality of
that action.
1.2 Assumptions of standard economics
● Structure of preferences over outcomes (axioms like transitivity, completeness etc.)
● Only outcomes matter (final wealth) not changes, not intentions
● Belief updating and subjective probability (doesn’t matter what the intention is)
● No (or negligible) cost for information or decision costs (effort and time)
● No limits of reasoning power
● No self control problems
1.3 Examples of violations of standard assumptions
1. consumer preferences
- Completeness: if we have two bundles A and B then either A is preferred to B, or B is
preferred to A, or the consumer is indifferent between A and B. “I don’t know” is not
allowed!
- Transitivity: if A is preferred to B and B is preferred to C than A is preferred to C.
Completeness: you know what you like
- If you have trouble to choose, you are apparently indifferent
- But that means that if a very little extra is added to one of the options, that option will
be preferred.
- If one of the suitors would tell her: “if you choose me I will buy you a bicycle”: would
that help?
- If individuals have trouble choosing,
they tend to:
● Postpone a decision if possible
● Stick to the current situation if possible
● Accept the default if available
Status Quo bias: stick to the current situation
Default bias: the tendency to make the choice for which no action has to be taken
Transitivity
A scores better than B on Time and Teacher
B scores better than C on Time and Content
But... C scores better than A on Teacher and Content
violation of transitivity →
Condorcet paradox In many cases the decision maker is not an individual but a group (e.g.
family, firm). In these cases Condorcet cycles (you stay in a cycle) can exist.
A committee with three members have to vote on three options; first two options and than
the winner against the third one. The outcome depends on the order of votes
If member 1 (point in time) is chairman: B versus C, A against winner: A wins
If member 2 (teacher) is chairman: A versus B, C against winner: C wins
If member 3 (content) is chairman: A versus C, B against winner: B wins
,why intransitivity is bad →
Money machine to extract money from decision makers with intransitive
preferences:
If A is preferred to B, B is preferred to C and C is preferred to A;
- The consumer possesses A.
- Offer: if you pay 1 cent you can change your A for C
- The consumer pays 1 cent and now possesses C
- Offer: if you pay 1 cent you can change your C for B
- The consumer pays 1 cent and now possesses B
- Offer: if you pay 1 cent you can change your B for A
- The consumer pays 1 cent and now possesses A
Result: the consumer loss 3 cents and again possesses A; we can repeat this until the
consumer is bankrupt or makes his preferences transitive!
In many situations outcomes are uncertain
Risk: probabilities are known (throw a 6 with a standard dice)
Ambiguity: probabilities are unknown (e.g. the Dow Jones will be >40000 on December 31
of this year)
Ambiguity is much more common than risk in real life
Standard Econ with risk: Expected Utility
Two more axioms needed:
- Continuity: If I prefer 4 apples over 4 pears, I should also prefer 2 apples and 2
pears over 4 pears. (If X is preferred over Y then aX+(1-a)Y will be preferred over Y
for all 0<a<1)
- Independence of irrelevant alternatives: If X ∼ Y (the decision-maker is indifferent
between X and Y) then for all Z and p [p:X , (1-p):Z] ∼ [p:Y , (1-p):Z]
(the decision-maker is indifferent between a lottery with prize X with probability p and
a prize Z with probability (1-p) and the lottery with prize Y with probability p and a
prize Z with probability (1-p)). The idea is that in both lotteries the probability (1-p) to
receive Z is the same, and the lotteries only differ in the prize that is received with
probability p (respectively X and Y) so only the preference for X and Y should
influence the preference between the two lotteries.
If these 4 (Completeness,Transitivity, continuity and Independence of irrelevant alternatives)
hold → there exists an Expected Utility function such that:
U(p:X , (1-p):Y)=pU(X) + (1-p)U(Y)
In EU two psychological properties
- Diminishing/increasing sensitivity for money
- Risk attitude risk averse/risk loving are the
same!
, Standard Econ with ambiguity: Subjective Expected Utility
- When the probabilities are not known standard economics assumes that the
decisionmaker attaches a subjective probability to each possible event.
- Needed assumptions:
● Subjective probabilities add up to not more than 1
● If A is a subset of B, then P(A)≤P(B)
● The probabilities are independent of the consequences of the event (no optimism or
● pessimism allowed!)
Violation of this assumptions:
❖ Independence of irrelevant alternatives in EU
- What do you prefer:
A: 30 euros with probability 1
B: 40 euros with probability .80, 0 with probability .20
- What do you prefer:
C: 30 euros with probability .25, 0 with probability .75
D: 40 euros with probability .20, 0 with probability .80
- Most prefer A over B, but D over C. But:
E: play lottery A with probability .25, 0 with probability .75
F: play lottery B with probability .25, 0 with probability .75
E=C and F=D !
❖ Optimism
the subjective probability is larger for events that are positive for the decision-maker. In
Western culture (especially North-American) optimism is seen as a positive personality
trait, while to economists it is considered a bias (which can lead to suboptimal decisions).
❖ Additivity biases
a) How likely is it that the Dow Jones will end December 31 between 40000 and 42000?
b) How likely is it that the Dow Jones will end December 31 between 42000 and 45000?
c) How likely is it that the Dow Jones will end December 31 between 40000 and 45000?
Typically a+b>c
❖ Availability
- The subjective probability of an event is in practice based upon how easy it is to
imagine the event or to retrieve past instances
- Ajax plays against an amateur club. How likely is it that:
1. Ajax will get 1-0 behind
2. Ajax will get 1-0 behind, but in the end will win
- Pr(2) should be smaller than Pr(1), but Pr(2) could be estimated more
likely than Pr(1)
, 1.4 bounded rationality
- Standard theory assumes that information is costless, information processing is
costless, the agent optimizes and make no errors
- In reality people make errors (and they know that), decision costs can be large
compared with the cost of errors and many problems are to complicated. Therefore
agents simplify problems and look for a solution that is good enough.
Rational decision making “simple” case (no uncertainty)
- Alternatives: What are the possible actions I can take and to what outcomes do they
lead?
- criterions: What are the criterions to compare the outcomes?
- Relative importance of criterions: What is the relative importance of each criterion?
- Calculate: Analyze each action/outcome, using weighted criterions
- Choose: Choose the best action
the assumption that we know our preferences is not realistic, Do we have to find out all
relevant characteristics for all these houses? Too many options → simplify the problem:
filtering alternatives, formulating a minimum requirement. (fewer criterions, fewer
alternatives)
Herbert A. Simon Started with studying decision processes within firms:
1. Many problems are too complex, or not worthwhile
2. Firms searches for a solution that is “good enough”
3. Firms simplify the problem by formulating sub-goals
4. Use of heuristics and tradition (how did we do it in the past, and can we use that)
5. Change behavior only if you are not happy with the outcomes, search locally for
improvement. (Only when a problem is detected, the search for a solution will start)
6. Decision tasks are distributed, coordination is by communication and hierarchy
The first 5 points are also relevant for individual decision makers
Intuition and learning
How you simplify the problem and which action you consider depends on characteristics
of the situation.
- An expert is much better in this than a starter → But also an expert cannot tell you
how he does this
- Intuition is based upon experience
- Learning is easier with direct (time) and univocal (good or bad, easy to feedback
does not really change) feedback
- A problem is that learning by experience is asymmetric → The decision maker
chooses an action, gets feedback and improves his knowledge about the quality of
that action.
