CAPM and APT

Lakehead University

Winter 2005

Assumptions under the CAPM

All investors are risk-averse individuals maximizing the expected utility of their wealth. Investors are price takers, have homogeneous expectations and returns are normally distributed. There exists a risk-free rate at which investors can lend and borrow unlimited amounts.

Assumptions under the CAPM

The quantities of assets are xed and assets are perfectly divisible. Asset markets are frictionless, information is costless and simultaneously available to all investors. There are no taxes, regulations or restrictions on short selling.

E[R] T
m

Capital Market Line (CML)

rf

e u e
Efcient Frontier

Efciency of the Market Portfolio

The market portfolio (m) lies on the efcient frontier. The market portfolio contains all existing assets, the weight of asset i in m being wi = market value of asset i . market value of all assets

The Security Market Line

It can be shown that
2 m = w2 2 + w2 2 + . . . + w2 2 + 2w1 w2 12 + 2w1 w3 13 + . . . 1 1 2 2 n n = w1 1m + w2 2m + . . .

where im = COV(Ri , Rm ). This means that the contribution of each asset to the risk of the market portfolio depends on the covariance of the asset return with the market portfolio return.

The Security Market Line

Therefore, in equilibrium, the excess return of each asset (or portfolio of assets) over the risk-free rate per unit of covariance risk must be the same for all assets, i.e. E[Ri ] r f E[R j ] r f = im jm for all i, j.

The Security Market Line

This must also be true with the market portfolio and thus E[Ri ] r f E[Rm ] r f E[Rm ] r f = = im mm 2 m for all i, which gives E[Ri ] = r f + im E[Rm ] r f . 2 m

The Security Market Line

We now dene a term, beta (), as i = im im i = . 2 m m

The coefcient i measures the systematic risk of asset i.

The Security Market Line

Equation E[Ri ] = r f + i E[Rm ] r f . is referred to as the security market line.

10

E[R] T

Security Market Line (SML)

E[Rm ] r f

E[Rm ]

rf

11

The Security Market Line

An asset below the SML is overvalued. An asset above the SML is undervalued.

12

E[R] T

x (undervalued)

Security Market Line (SML)

rf

y (overvalued)

13

Properties of the CAPM

Total risk = systematic risk + unsystematic risk: Ri = ai + bi Rm + i and thus 2 = b2 2 + 2i . i i m

14

Properties of the CAPM

Under the CAPM, E[Ri ] r f i
Sharpe ratio of Asset i

= im

E[Rm ] r f m
Sharpe ratio of m

15

Properties of the CAPM

If portfolio x has a beta of x and portfolio y has a beta of y , then a portfolio p containing a fraction a of x and a fraction 1 a of y has a beta of p = ax + (1 a)y .

16

Extensions of the CAPM

No Riskless Asset How does the model change when there is no riskless asset? Let z denote the portfolio on the minimum-variance frontier (not the efcient set) that has a covariance of zero with the market portfolio. Then all assets can be priced as follows: E[Ri ] = E[Rz ] + i E[Rm E[Rz ] , where i =
im . 2 m

17

Empirical Tests of the CAPM

To test the CAPM, securities betas are rst calculated, portfolios are then formed and the model R pt = 0 + 1 p + pt is estimated, where R pt = R pt r f t .

18

Empirical Tests of the CAPM

If the CAPM holds, the estimate of the parameter 0 should not be signicantly different from zero; the estimate of the parameter 1 should be equal to Rmt r f t , the excess return of the market portfolio, the market risk premium.

19

Empirical Tests of the CAPM

These tests have shown that: The intercept 0 is signicantly different from zero and the slope 1 is smaller than the market risk premium. Factors other than beta are successful in explaining the portion of security returns not captured by beta. Fama and French (1992) is one of the most cited studies testing the CAPM.

20

Empirical Tests of the CAPM

Fama and French (1996) propose a three-factor model: E[Ri ] r f = i E[Rm ] r f + si E[SMB] + hi E[HML]

21

CAPM and the Cost of Equity

The CAPM is often used to calculate a rms cost of equity. When doing so, one has to determine: A risk premium The companys beta A risk-free rate

22

CAPM and the Cost of Equity

Risk Premium The cost of equity, kE , is a forward-looking cost of equity. That is, it is based on what will happen rather than what has happened. Its calculation should involve a forward-looking risk premium.

23

CAPM and the Cost of Equity

Risk Premium According to Arnott and Bernstein (2002), the forward-looking risk premium is around 0-2%. Computing a forward-looking risk premium can be quite cumbersome and is very subjective. An average of past risk premia is thus often used as a proxy. What average to use for the risk premium? An arithmetic average? A geometric average? Both have advantages and disadvantages.
24

CAPM and the Cost of Equity

Risk Premium Based on Ibbotson Associates data from 1926 to 1995, the equity risk premium is as follows:
Equity Market Risk Premium (km kf ) k f = T-Bill Return Arithmetic Mean Geometric Mean 8.5% 6.5% k f = T-Bond Return 7.0% 5.4%

25

CAPM and the Cost of Equity

Beta Which to use? Calculate your own: Run the regression kE,t = + km,t . Use beta from a published source (Bloomberg, Value Line, Standard & Poors, etc.)

26

CAPM and the Cost of Equity

Beta When calculating your own , what type of data to use? Monthly returns? Weekly returns? Increasing the number of period increases the reliability of the estimate but incorporates information that may not be irrelevant. Reducing the observation period from monthly to weekly may increase the number of observations but may yield observations that are not normally distributed.

27

CAPM and the Cost of Equity

Risk-Free Rate What to use for k f ? Textbooks often suggest to use the 90-day T-bill rate. The life of an investment is usually more than 90 days. Firms mostly use yields on government bonds that match with the life of their investments (10-year government bonds, for instance).

28

Arbitrage Pricing Theory

The arbitrage pricing theory (APT) is based on the law of one price, which states that two stock with the same risk characteristics must sell at the same price. Market participants develop expectations about factors that inuence stock returns and realized stock returns depend on the difference between the expected and realized values of these factors.

29

Arbitrage Pricing Theory

Suppose asset returns are randomly generated according to an n-factor model: Rit = E [Rit ] + bi1 f1t + bi2 f2t + . . . + bin fnt + eit where fkt is the deviation of the realized value of factor k from its expected value.

30

Arbitrage Pricing Theory

GDP per capita, for example, could be a factor. If FGDP denotes the risk premium associated with the GDP, then the expected return in stock i attributed to per capita GDP at time t is bi,GDP E [FGDP,t ]

31

Arbitrage Pricing Theory

If we can nd a risk premium for all n factors, the expected return on security i at time t is E [Rit ] = a0 + bi1 E [F1t ] + bi2 E [F2t ] + . . . + bin E [Fnt ] and fkt is given by fkt = Fkt E [Fkt ] , k = 1, 2, . . . , n

32

Arbitrage Pricing Theory

Let k denote the expected return of a portfolio with unit sensitivity with factor k and zero sensitivity with the other factors, k = 1, . . . , n. Then, for all i,
E[Ri ] = r f + i1 1 r f where, for k = 1, . . . , n, ik = Cov(Ri , k ) Var(k ) . + i2 2 r f + . . . + in n r f ,

33