FINANCIAL ECONOMICS
Topic 3: Theory of choice: Utility Theory
Given Uncertainty
Risk aversion measurement
• For the second risk in the above example
• Markowitz risk premium: $489 = expected wealth – CE =17,200-16,711
• Pratt-Arrow risk premium: $324
• Markowitz measure of risk premium is more preffered for large or
asymmetric risks.
• Pratt-Arrow risk aversion measure is useful for distinguishing between
different types of concave utility functions.
(Pratt-Arrow có ý nghĩa trong việc xác định hàm lợi ích hợp lý)
Stochastic Dominance
So sánh tài sản bằng xác suất
• First-order stochastic dominance (So sánh bậc 1):
• An asset (or portfolio) is stochastically dominant over
another if an individual receives greater wealth from it ini
every (ordered) state of nature.
• X is stochastically dominant over y if:
𝐹𝑥 𝑊 ≤ 𝐺𝑦 𝑊 for all 𝑊.
𝐹𝑥 𝑊 < 𝐺𝑦 𝑊 𝑓𝑜𝑟 𝑠𝑜𝑚𝑒 𝑊𝑖
• 𝐹𝑥 𝑊 , 𝐺𝑦 𝑊 :Cumulative probability Distribution Fuction)
defined on wealth 𝑊
• assuming
• Fx(W) and Gx(W) are increasing utility functions.
• distributrion of wealth provided by both x and y is normal distribution
Stochastic Dominance
• Risk adverse, neutral and lover all
prefer 𝑥 to 𝑦 because of 𝐸[𝑈 𝑊 ] 𝑥>
𝐸[𝑈 𝑊 ] 𝑦 for every increasing utility
function.
where 𝑓(𝑊) : frequency distribution of
wealth
(xác suất của một trạng thái của nền kinh tế.
• The opposite for nonincreasing utility
functions.
Stochastic Dominance
• 2nd order stochastic dominance:
• Assumtions: utility functions are
nondecreasing and strictly concave
(meaning investors are risk averse)
• Asset x will be stochastically dominant
over asset y if:
Stochastic Dominance
• 2nd order stochastic dominance:
• Because the utility function is concave,
∆𝑈1 > ∆𝑈2
• If utility curve is linear, ∆𝑈1 = ∆𝑈2 , neutral
investors take 𝑥 = 𝑦
• Because cumulative distribution
𝐶𝐷𝐹(𝑦−𝑥) > 0 for all W , so x> y.
Using mean and variance as choice criteria
• Using 𝜇, 𝜎
• 𝑋~𝑁 𝜇1 , 𝜎1 , 𝑌~𝑁 𝜇2 , 𝜎2 → 𝑍 = 𝛼𝑋 + 𝛽𝑌 ~𝑁 𝜇3 , 𝜎3
• If X, Y both have normal distribution, the linear combination of X
and Y (portfolio of X and Y) also has normal distribution.
• 𝜇, 𝜎 are 2 moments of a normal distribution.
• There is also skewness and kurtosis
• 𝜇 is mean, 𝜎 is the risk of asset
Using mean and variance as choice criteria
• Within one period:
ഥ 𝜎𝑊 ) , then according to distribution rule
If 𝑊~𝑁(𝑊,
𝐸 𝑊𝑗 2
𝜎𝑊
෪
𝑅𝑗 𝑖𝑠 𝑎 𝑛𝑜𝑟𝑚𝑎𝑙 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑤𝑖𝑡ℎ 𝐸 𝑅𝑗 = − 1 and 𝜎𝑅2 =
𝑊0 𝑊02
• Explanation:
𝜎𝑅2 = 𝐸[𝑅𝑗 − 𝐸 𝑅𝑗 ]2
෪𝑗
𝑊 𝐸 𝑊𝑗
= 𝐸[ −1− + 1]2
𝑊0 𝑊0
2
1 𝜎𝑊
෪𝑗 − 𝐸 𝑊𝑗 ]2 =
= 2 𝐸[𝑊
𝑊0 𝑊02
Using mean and variance as choice criteria
• Assuming asset returns are normally distributed, with
mean 𝐸 and standard deviation 𝜎, utility function will
be: (Tobin (1958):
𝑈 = 𝑈 𝑅𝑗 , 𝐸, 𝜎
• Expectred utility
• Where
𝑓(𝑅; E; 𝜎) là is distribution of 𝑅
Using mean and variance as choice criteria
• Relationship between return and risk
for risk averter:
• Expressed through indifference curves:
set of investment opportunities with the
same 𝐸(𝑈)
• That indifference curve is increasing and
concave (second derivative >0)
• Only applied for risk averter
Indifference curve of risky investments
• Prove that indifference curve is increasing
function
• Change from normal distribution to (0, 1) distribution:
෨
𝑅~𝑁 ෨
𝜇, σ → 𝑍~𝑁(0,1)
Indifference curve of risky investments
• Because investors are indifferent among investment opportunities:
∞ ∞
𝑑𝐸
0= න 𝑈 ′ (𝐸 + 𝜎𝑍)f Z; 0; 1 dZ + න 𝑈 ′ (𝐸 + 𝜎𝑍)Zf Z; 0; 1 dZ
𝑑𝜎
−∞ −∞
∞
𝑑𝐸 −∞ 𝑈 ′ (𝐸+𝜎𝑍)Zf Z;0;1 dZ
=− ∞ > 0 ??
𝑑𝜎 −∞ 𝑈 ′ (𝐸+𝜎𝑍)f Z;0;1 dZ
Indifference curve of risky investments
∞
𝑑𝐸 −∞ 𝑈 ′ (𝐸+𝜎𝑍)Zf Z;0;1 dZ
=− ∞ >0
𝑑𝜎 −∞ 𝑈 ′ (𝐸+𝜎𝑍)f Z;0;1 dZ
• Denominator >0
bec𝑎𝑢𝑠𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑎𝑠𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 𝑡ℎ𝑎𝑡 𝑚𝑎𝑟𝑔𝑖𝑛𝑎𝑙 𝑢𝑡𝑖𝑙𝑖𝑡𝑦
𝑈 ′ (𝐸 + 𝜎𝑍) > 0
• Nominator <0 because
“marginal utility when 𝑍 < 0” > “marginal utility
when 𝑍 > 0"
→ Increasing function
Mean – Variance paradox
Hạn chế khi sử dụng phương pháp mean-variance
State of economy
Recessio Bad Normal Good Boom
n
Operating profit
Probability
Firm A
Interest
Before tax profit
Tax 50%
Net income
EPS (200 shares)
Firm B
Interest
Before tax profit
Tax 50%
Net income
EPS (100 shares)
($2.82, $7)
($1.41, $5)
Risk aversion level 𝐼𝐼𝐼 > 𝐼 > 𝐼𝐼
Firm A
Firm B
Mean-variance paradox
• Graph shows A is preferred to B
• Mean-Variance → B is preferred to A
• Mean-Variance assumes normal distribution , while EPS is
non-normal distribution in this case
Tổng kết nội dung chính
• 5 tiên đề về hành vi của các cá nhân khi đầu tư (NĐT)
• Sử dụng tiên đề để xây dựng hàm lợi ích của nhà đầu tư
• NĐT lựa chọn dựa trên tiêu chí tối đa hóa ích lợi kỳ vọng
(Expected Utility)
• Đo lường mức độ sợ rủi ro của NĐT (risk premium)
• Phương pháp Markowitz
• Phương pháp Pratt-Arrow
• So sánh tài sản
• So sánh bậc 1
• So sánh bậc 2
• Đường bàng quan của NĐT trong không gian mean-variance