Economics cheatsheet
Reference-dependent utility function
Choices with risk
Prospect = list of probabilities that things will happen together with monetary payoff
the list: ( p1 , x1 ; p2 , x2 ; …; pn , xn ),
where p i is the probability of getting monetary payoff x i.
Prospect X = (0.5, $0; 0.5, $100), in which Alan has a 50:50 chance of winning $100.
Expected
utility = a
model of how
people
choose
prospects involving risk
Assumptions
1. Transitivity of preferences (if X ≥ Y and Y ≥ Z then X ≥ Z)
2. Independence of irrelevant alternatives (sth irrelevant, shouldn’t matter)
3. Completeness of preferences (never indifferent – choices have to be made)
4. Conitnuity
Rank expected utility (REU) = perception of probabilities
A function that weighs probabilities
where gamma is the degree of misperception,
always between 0 and 1,
1 indicating no misperception so EU can be used
with a parameter of 0.1, 10% probability is perceived as 14.4%
When Gamma < 1,
smaller P’s are overweighted,
higher P’s underweighted
Coin toss:
(0.5, 0$; 0.5, 100$)
Allais paradox
Individuals prefer certainty over a risky outcome even if this defies the expected utility
axiom. People favor outcomes that are perceived as certain rather than probable or
possible. This is called the certainty effect. (= γ, δ < 1)
The Allais paradox could be explained by pessimism and the overweighting of a
small probability of getting $0.
Exponential discounting
An exponential curve has a constant discount rate.
Would you prefer $100 today or $150 in a year’s time?’.
If they answer ‘$100 today’ then the discount factor is smaller than 0.66
discount factor
(the higher, the more impatient the person becomes)
leads to changes in utility based on the discount factor
captures time preference
(patience = close to 1, impatience = close to 0
discounting can also be expressed by the discount rate
If discount factor is 0.79, then the discount rate is 0.27.
This means an interest rate of 27% is required to delay receiving $100
If the discount factor is 0.8 then ‘$10 worth of utility’ next period is equivalent to $8
today.
More generally, $10 next period is equivalent to δ $10 today.
Exponential discounting indifference
Let’s convert monetary amounts into utilities: 100$ vs 160$ in 1 year
a = 100^0.5 = 10; b = 160^0.5 = 12.65
Given you are indifferent between a and b at time zero, we know that:
• U0(a) = U0(b)
• which implies that 10 = 12.65δ
• which is to say that δ = 10/12.65 = 0.79
Absolute magnitude effects
the discount factor is larger for larger amounts of money
people are more patient for larger amounts
Hyperbolic discounting decreasing impatiene
Individuals may be willing to wait for a
reward that is further away in time but tend to
be impatient for shorter-term rewards
gives a higher discount factor for longer
periods
CHOICES NEED TO BE CONSISTENT OVER TIME
a hyperbolic discount curve has a higher discount rate in the near future and a lower
discount rate in the distant future.
Quasi hyperbolic discounting
Can explain time inconsistent preferences
Give less weight to future utility
People tend to be patient for long-term gains, but impatient for short-term gains:
100$ in 30 days or 110$ in 31 days
The ratio between discounting factors tells what is 100$ today compared to 110$ tmr:
110$ tmr = 110 x 0.99 = 108.9
108.9 > 100 so choose 110$
β is the present bias when β < 1 more weight is given to today than the future
=there are present-biased preferences and decreasing impatience
always remember periods!
t0 = u, T1 = Beta x Gamma, T2 = Beta x Gamma^2…
Statistical analysis
Chi-square test of a contingency table tests if two variables are related.
1. Number of cases expected if nominal variables are independent = H0
P (BA) x P (Finance) x total n (60/152) x (61/152) x 152 = 24.0789
2. Compute the expected value for all cells
3. Calculate test statistic:
4. Reject the 0 hypotheses if test statistic (x^2) > than the critical value
5. Degrees of freedom by deleting 1 row & 1 column how many values (x)?
6. Compare test statistic value with probability distribution table for value (x)
7. If test statistic value > probability distribution value for n degrees of freedom
reject H0
Utility of a sequence (loss aversion)
If β < 0 preference for decreasing sequence
If β > 0 preference for improving sequence
If σ < 0 preference for a smooth sequence (closer values for AU / RU)
If σ > 0 preference for one-sided sequence
= 0.5(AU-RU)
FR: AU = 0x-5 + 1x5 + 2x10 + 3x4 = 37
RU = 3x-5 + 2x5 + 1x 10 + 0x4 = 5
Time & Risk
Choose the money today if: c+100 + δc > c + δ(c+110) risk neutral
c is the amount she would normally consume in a day and δ is her discount factor.
drop c for calculation
If risk averse, the person is more inclined to take the money today.
Compare function to a higher amount of $:
