16-Port Selection Mean Var Page 489 Wednesday, February 4, 2004 1:09 PM
Portfolio Selection Using Mean-Variance Analysis 489
rA()x = –U′′()x ⁄U′() x
This measure expresses the intuitive fact that the more the utility func-
tion is curved, the more the investor is risk-averse. Listed below are
some examples of utility functions:
■ Linear utility function:
Ux()= abx + , U′()x = b , U′′()x = 0
The linear function is not concave; it represents a risk-neutral investor.
■ Power utility function:
1 – a
Ux()= ---------------------x –^1 - , U′()x = x –a , U′′()x = –ax – a –^1 < 0
1 – a
The power utility function is concave; it represents a risk-averse investor.
■ Logarithmic utility function:
Ux()= ln ()x , U′()x = 1 ⁄x , U′′()x = – 1 ⁄x^2 < 0
The logarithmic utility function is concave; it represents a risk-averse
investor.
■ Quadratic utility function:
Ux()= abx + – ---x^2 , U′()= bcx , U′′()= –c < 0
c
x – x
2
The quadratic utility function is concave; it represents a risk-averse
investor.
In a probabilistic setting, the utility function is a monotone function
of a random variable and is, therefore, a random variable itself. To opti-
mize, one single utility number must be defined for each portfolio
choice. Utility is therefore defined as the expected value of stochastic
utility: