Derivation of the Capital Asset Pricing

Model (CAPM)
The CAPM begins with a specific utility function, where the utility of a random distribution of
wealth depends solely on the first two moments of the probability distribution: the mean and
the variance.

It’s important to note that this is only compatible with the expected utility model when all
assets are normally distributed or when the expected utility function is quadratic. Nonetheless,
mean-variance can serve as an approximate representation of the general utility function in a
broader range of scenarios. In this context, "risk aversion" implies that an increase in expected
consumption is desirable, while an increase in consumption variance is undesirable.

We start by establishing the budget constraint in an intertemporal model, which spans two
periods, 𝑡𝑡0 and 𝑡𝑡1 . We derive the investor second-period consumption, which describes how
consumption is linked to wealth (W), consumption in 𝑡𝑡0 (c), and rate of return:
𝐴𝐴 𝐴𝐴

𝐶𝐶𝑡𝑡1 = (𝑊𝑊 − 𝐶𝐶𝑡𝑡0 ) � 𝑥𝑥𝑎𝑎 𝑅𝑅𝑎𝑎 = (𝑊𝑊 − 𝐶𝐶𝑡𝑡0 ) �𝑥𝑥0 𝑅𝑅0 � 𝑥𝑥𝑎𝑎 𝑅𝑅𝑎𝑎 �
𝑎𝑎=0 𝑎𝑎=1

Where:

𝐶𝐶𝑡𝑡1 : investor' s consumption in period 1

W: wealth of investor in period 0

𝐶𝐶𝑡𝑡0 : investor' s consumption in period 0

𝑅𝑅𝑎𝑎 : total return on asset a regarded as a random variable

𝑥𝑥𝑎𝑎 : amount purchased of asset a

𝑥𝑥0 : amount purchased of risk-free asset

𝑅𝑅0 : total return on a risk-free asset

Since the portfolio weights must sum to one, so that 𝑥𝑥0 = 1 − ∑𝐴𝐴𝑎𝑎=1 𝑥𝑥𝑎𝑎 , we can rewrite the
budget constraint as:
𝐴𝐴

𝐶𝐶𝑡𝑡1 = (𝑊𝑊 − 𝐶𝐶𝑡𝑡0 ) �𝑅𝑅0 + � 𝑥𝑥𝑎𝑎 (𝑅𝑅𝑎𝑎 − 𝑅𝑅0 )�

𝑎𝑎=1

By Ivan Pardo Gonzalez

https://www.linkedin.com/in/ivan-pardo-gonzalez/
The expression in square bracket is the portfolio return. The consumption in period 1, depends
on how much money the investor saved (𝑊𝑊 − 𝐶𝐶𝑡𝑡0 ) in 𝑡𝑡0 and the rate of return on this
savings/investment.

Given the assumptions about mean-variance utility function, whatever the level of investment,
the investor would like to have the least possible variance of the portfolio for a given expected
value. That is, the investor would like to purchase a portfolio that is mean-variance efficient.

Please keep in mind that the portfolio chosen will depend on the investor’s utility function
(indifference curve), but whatever it is, it must minimize variance for a given level of expected
return.

Step 1: Problem Setup

We start with the mean-variance portfolio optimization problem. The goal is to minimize the
variance of the portfolio return subject to the constraints that we achieve a specified expected
return 𝑅𝑅�, and we satisfy the budget constraint ∑𝐴𝐴𝑎𝑎=0 𝑥𝑥𝑎𝑎 = 1

The optimization problem is:

𝐴𝐴 𝐴𝐴

𝑚𝑚𝑚𝑚𝑚𝑚𝑥𝑥0 ,…,𝑥𝑥𝐴𝐴 � � 𝑥𝑥𝑎𝑎 𝑥𝑥𝑏𝑏 𝜎𝜎𝑎𝑎𝑎𝑎

𝑎𝑎=0 𝑏𝑏=0

Where:

𝑥𝑥𝑎𝑎 : amount purchased of asset a

𝑥𝑥𝑏𝑏 : amount purchased of asset b

𝜎𝜎𝑎𝑎𝑎𝑎 = 𝑐𝑐𝑐𝑐𝑐𝑐 (𝑅𝑅𝑎𝑎 , 𝑅𝑅𝑏𝑏 ) - the covariance between the returns on assets a and b

Subject to the constraints:

1. Expected return constraint:

𝐴𝐴

� 𝑥𝑥𝑎𝑎 𝑅𝑅�𝑎𝑎 = 𝑅𝑅�

𝑎𝑎=0

2. Budget constraint:
𝐴𝐴

� 𝑥𝑥𝑎𝑎 = 1
𝑎𝑎=0

By Ivan Pardo Gonzalez

https://www.linkedin.com/in/ivan-pardo-gonzalez/
Step 2: First-Order Conditions
To solve this constrained optimization, we introduce two Lagrange multipliers:

• λ for the expected return constraint,

• μ for the budget constraint.

The Lagrangian function becomes:

𝐴𝐴 𝐴𝐴 𝐴𝐴 𝐴𝐴

ℒ(𝑥𝑥0 , … . . , 𝑥𝑥𝐴𝐴 , λ, 𝜇𝜇) = � � 𝑥𝑥𝑎𝑎 𝑥𝑥𝑏𝑏 𝜎𝜎𝑎𝑎𝑎𝑎 − λ �� 𝑥𝑥𝑎𝑎 𝑅𝑅�𝑎𝑎 − 𝑅𝑅� � − 𝜇𝜇 �� 𝑥𝑥𝑎𝑎 − 1�
𝑎𝑎=0 𝑏𝑏=0 𝑎𝑎=0 𝑎𝑎=0

Note that in this problem 𝑥𝑥𝑎𝑎 can be positive or negative. This means that the investor/consumer
can hold a long or a short position in any asset, including the riskless asset.

The first-order conditions take the form:

𝐴𝐴
���𝑎𝑎� − 𝜇𝜇 = 0
2 � 𝑥𝑥𝑏𝑏 𝜎𝜎𝑎𝑎𝑎𝑎 − λ𝑅𝑅
𝑏𝑏=0

For a=0,….,A

Since the objective function is convex and the constraint are linear, the second-order conditions
are automatically satisfied.

Step 3: Special Case for Risk-Free Asset and Risky Portfolio

Now, let (𝑥𝑥1𝑒𝑒 , … . . , 𝑥𝑥𝐴𝐴𝑒𝑒 ) be some portfolio consisting entirely of risky asses that is known to be
mean-variance efficient. Let’s suppose that one of the risky assets available to the investors
(asset e) is a mutual fund that holds this efficient portfolio (𝑥𝑥𝑎𝑎𝑒𝑒 ). Then, the portfolio that invest 0
in every asset except for asset e and 1 in asset e is mean-variance efficient. This means that such
a portfolio must satisfy the conditions given in equation above for each asset a=0,…,A.

Noting that for this portfolio 𝑥𝑥𝑏𝑏 = 0 for b ≠ e, we see that the 𝑎𝑎𝑡𝑡ℎ first -order condition
becomes:

���𝑎𝑎� − 𝜇𝜇 = 0
2𝜎𝜎𝑎𝑎𝑎𝑎 − λ𝑅𝑅

We now have two special cases:

1. For the risk-free asset a=0:

− λ𝑅𝑅0 − 𝜇𝜇 = 0

2. For the efficient portfolio a=e:

���𝑒𝑒 − 𝜇𝜇 = 0
2𝜎𝜎𝑒𝑒𝑒𝑒 − λ𝑅𝑅

By Ivan Pardo Gonzalez

https://www.linkedin.com/in/ivan-pardo-gonzalez/
When a=0, 𝜎𝜎𝑎𝑎𝑎𝑎 is zero since asset 0 is not risky. When a=e, 𝜎𝜎𝑎𝑎𝑎𝑎 = 𝜎𝜎𝑒𝑒𝑒𝑒 since the covariance of a
variable with itself is simply the variance of the random variable.

Step 4: Solving for λ and μ

From the first condition for a=0, we solve for μ:

���0
𝜇𝜇 = − λ𝑅𝑅

Substitute this into the second condition for a=e:

2𝜎𝜎𝑒𝑒𝑒𝑒 − λ𝑅𝑅𝑒𝑒 + λ𝑅𝑅0 = 0

which simplifies to:

2𝜎𝜎𝑒𝑒𝑒𝑒 = λ(𝑅𝑅𝑒𝑒 − 𝑅𝑅0 )

Solving for λ

2𝜎𝜎𝑒𝑒𝑒𝑒
λ=
𝑅𝑅𝑒𝑒 − 𝑅𝑅0

Now, using 𝜇𝜇 = −λ𝑅𝑅0 , substitute λ back into the expression for μ

2𝜎𝜎𝑒𝑒𝑒𝑒 𝑅𝑅0
𝜇𝜇 = −
𝑅𝑅𝑒𝑒 − 𝑅𝑅0

Step 5: Substituting Back into the First-Order Condition

We now substitute the values of λ and μ back into the first-order condition:

���𝑎𝑎� − 𝜇𝜇 = 0
2𝜎𝜎𝑎𝑎𝑎𝑎 − λ𝑅𝑅
2𝜎𝜎𝑒𝑒𝑒𝑒 2𝜎𝜎𝑒𝑒𝑒𝑒 𝑅𝑅0
Substituting λ = (𝑅𝑅 and 𝜇𝜇 = :
𝑒𝑒 +𝑅𝑅0 ) 𝑅𝑅𝑒𝑒 −𝑅𝑅0

2𝜎𝜎𝑒𝑒𝑒𝑒 𝑅𝑅𝑎𝑎 2𝜎𝜎𝑒𝑒𝑒𝑒 𝑅𝑅0

2𝜎𝜎𝑎𝑎𝑎𝑎 − + =0
𝑅𝑅𝑒𝑒 − 𝑅𝑅0 𝑅𝑅𝑒𝑒 − 𝑅𝑅0

Dividing through by 2:
𝜎𝜎𝑒𝑒𝑒𝑒
𝜎𝜎𝑎𝑎𝑎𝑎 = (𝑅𝑅 − 𝑅𝑅0 )
𝑅𝑅𝑒𝑒 − 𝑅𝑅0 𝑎𝑎

By Ivan Pardo Gonzalez

https://www.linkedin.com/in/ivan-pardo-gonzalez/
Step 6: Rearranging to Derive CAPM
Rearranging the final equation, we get:
𝜎𝜎𝑎𝑎𝑎𝑎
𝑅𝑅𝑎𝑎 = 𝑅𝑅0 + (𝑅𝑅 − 𝑅𝑅0 )
𝜎𝜎𝑒𝑒𝑒𝑒 𝑒𝑒

This is the Capital Asset Pricing Model (CAPM) equation, where:

𝑅𝑅𝑎𝑎 is the expected return on asset a

𝑅𝑅0 is the risk-free rate

𝑅𝑅𝑒𝑒 is the return on the efficient portfolio

𝜎𝜎𝑎𝑎𝑎𝑎
is the beta or systematic risk or theoretical regression coefficient
𝜎𝜎𝑒𝑒𝑒𝑒

𝜎𝜎𝑎𝑎𝑎𝑎 is the covariance of returns of asset a and the returns of the efficient portfolio

𝜎𝜎𝑒𝑒𝑒𝑒 is the variance of the returns of the efficient portfolio/market

Finally, let’s dive into the beta

• Beta measures the sensitivity of an asset’s return to changes in the market return
(efficient portfolio).
• Beta is a semi-elasticity; it measures the change in an asset's expected return for a unit
change in the market return. It does not measure a proportional change (i.e.,
percentage change) in the asset's return with respect to the market return, but rather
the absolute change in return.
For example:
If the beta of an asset is 1.5, this means that for every 1% increase in the market
portfolio's return, the asset’s return will increase by 1.5% on average. If the market
return falls by 1%, the asset’s return would decrease by 1.5%.
• Beta is the slope of the security characteristic line (SCL), which plots the asset's excess
returns against the market's excess returns. The SCL is given by the following regression
equation:
𝑅𝑅𝑎𝑎 = 𝛼𝛼 + 𝛽𝛽𝑅𝑅𝑒𝑒 + 𝜀𝜀
Where:
𝑅𝑅𝑎𝑎 is the return of the individual asset
𝑅𝑅𝑚𝑚 is the return of the market or efficient portfolio
𝛼𝛼 is the intercept, also called alpha, representing the asset’s return independent of the
market' s return
𝛽𝛽 is the slope coefficient

By Ivan Pardo Gonzalez

https://www.linkedin.com/in/ivan-pardo-gonzalez/
 𝜀𝜀 is the error term, which represents the idiosyncratic component of the asset’s return
not explained by the market.

Sources:

• Varian Hal, Microeconomic Analysis.

• Eugene F. Fama and Kenneth R. French, The Capital Asset Pricing Model: Theory and
Evidence.
• Edward J. Sullivan, Lebanon Valley College, A Brief History Of The Capital Asset Pricing
Model

By Ivan Pardo Gonzalez

https://www.linkedin.com/in/ivan-pardo-gonzalez/