FINANCE 361 – Topic 8 – CAPM and APT

Paul Geertsema

1
Contents

1 Readings 4

2 What are we doing today 5

3 CAPM Introduction 6

4 My view of the CAPM (and other theories) 8

5 CAPM Assumptions (all 10) 10

6 CAPM development 11

7 Some properties of the CAPM 14

8 Understanding the CAPM 15

9 Expected return 16

10 Risk free rate 17

2 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 Contents (cont.)

11 Market price of risk 18

12 Amount of risk 19

13 Worked example 21

14 Estimating the CAPM beta 22

15 Arbitrage Pricing Theory (APT) 27

16 CAPM vs APT 28

3 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
1 Readings

• Read BKM Ch 9, 10

4 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
2 What are we doing today

• Review the CAPM

• Introduce the APT framework
• Look at some common factor models

5 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
3 CAPM Introduction

• The CAPM is now attributed to Sharpe, Lintner and Mossin, building on the
work of Markowitz
• Core idea: What if

– Everybody have identical beliefs; and

– Everybody optimises using the Markowitz approach; and
– Nobody has enough wealth to move the market

• Result: CAPM (https://en.wikipedia.org/wiki/Capital_asset_pricing_model)

• Markowitz shows how a single investor should optimise a portfolio
• The CAPM shows what happens if everybody follows Markowitz

6 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 CAPM Introduction (cont.)

• The CAPM is, even today, the dominant theory of risk and return (in industry;
not in academia)

– In academia the Fama-French 5-factor model (2015 - ?) is the current

benchmark for empirical work
◦ But actively challenged by the Hou, Xue, Zhang (2014) q-factor model
◦ And a bunch of newer extensions, eg the Fama-French 6-factor model
and the HXZ 5-factor model.
– Supplanting the Fama-French 3-factor model and related Carhart 4-factor
model (1995 - 2015)
– Which superseded the 1-factor CAPM model (1975 - 1995)

• The CAPM model is not uncontroversial; arguments about it continues to rage

• The BKM take on the CAPM is, in my view, somewhat complacent

7 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
4 My view of the CAPM (and other theories)

• The simple CAPM is rejected by the data (realised returns)

• Even so, it is still useful, even if it is wrong

– But please remember that realised returns are also related to (many) other
things besides CAPM beta

• Finance it not a faith; you don’t have to believe things (like the CAPM)
• Finance is both a science and a craft

– The science demands that we remain sceptical and require evidence and
logically consistent reasoning
– The craft demands that we choose the right tool for the right job, and then
use it skilfully to solve the problem at hand
– Both science and craft demands understanding

• What is your view on the CAPM? Why?

• Again, the difference between normative and descriptive science

8 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 My view of the CAPM (and other theories) (cont.)

• CAPM fails descriptively (“how the world actually works”), but arguably has
more support normatively (“how you should make decisions”)
• Watch this space...

9 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
5 CAPM Assumptions (all 10)

1. Markets are

(a) not subject to transaction costs or taxes (friction-less markets)

(b) not influenced by individual investors (perfectly competitive)

2. Investors

(a) have access to every asset (integrated markets)

(b) are mean-variance optimisers (Markowitz)
(c) have identical beliefs (homogeneity)
(d) have the same single period horizon

3. Assets

(a) are infinitely divisible

(b) are tradable
(c) can be shorted without limit

4. Investors can invest in or borrow at a the same risk free rate (risk free asset)

10 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
6 CAPM development

• This explains the arguments used to develop the CAPM theory

– It is not a proof ...

– Search “CAPM derivation” if you are interested in the mathematical devel-
opment of the CAPM

11 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 CAPM development (cont.)

• Everybody holds the same risky portfolio because investors have

– Access to the same risky assets

– Mean-variance preferences
◦ Note: individual risk aversions can be different!
– Identical beliefs
– Identical horizons

12 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 CAPM development (cont.)

• Market risk is the only risk

– The market portfolio is the portfolio that contains all risky assets
– Everybody holds the same portfolio (see previous slide)
– Hence, everybody holds the market
– Only market risk matters

• Relying on Markowitz

– Asset returns should be linear in covariance of asset return with market

return, scaled by variance of market return
COV [rm ,ri ]
– V AR[rm ]

• E[ri] − rf = COV [rm ,ri ]

V AR[rm ] E[rm − rf ] = βE[rm − rf ]

• Remember, we use an estimated beta in practice βc ̸= β

13 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
7 Some properties of the CAPM

• Risk premium on market portfolio is proportional to average risk aversion times

the market variance
2
– E[rM ] − rf = ĀσM

• If an investor owns risky assets, he/she will own it in proportion to the market

– Investors allocate between risk free asset and market portfolio

– Implication: No need for different risky portfolios to cater to different investor
risk aversion profiles! (This is known as two fund separation)

14 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
8 Understanding the CAPM

• E[ri] − rf = βE[rm − rf ]
• E[ri] = rf + βE[rm − rf ]
• Expected return on asset i = risk free rate + (estimated amount of risk in asset
i) x (expected market price of risk)
• Let me break it down...

15 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
9 Expected return

• Probability weighted average of all possible future outcomes

• In CAPM everybody has the same beliefs

– Everybody agrees on the expected returns for all assets

• Expected returns differ from both realised future returns and realised past returns

– Sometimes we implicitly assume expected future returns = average of past

realised returns
– It is an assumption!

16 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
10 Risk free rate

• What is the risk free rate?

• Commonly assumed to be short term government debt (like T-bills)

– Is government debt really risk free?

• In the CAPM, the risk free rate should have a beta of zero

– Does government debt have zero beta?

• In recent data, government beta is sometime positive (ie Spain during sovereign
debt crises) and sometimes negative (US during GFC)
• If you are using this model in the “real world” you should think carefully about
the risk free rate you use

– Estimate its Beta!

– If it is not zero ... think why. Is it appropriate?

• Motivates the use of zero beta portfolios as a substitute for risk free asset (see
“Black extension to CAPM”)

17 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
11 Market price of risk

• Better known as the “market risk premium”: The expected return on holding
the market portfolio less the expected return on the risk free rate
• Note, my formulation of CAPM is a little different from the textbook version

– E[ri] = rf + βE[rm − rf ] vs E[ri] = rf + β(rm − rf )

– Can you spot the difference?

• I take the expectation of (rm − rf )

– Why? because the future market return is not known

– Even the future risk free rate may vary (it certainly has in the past)

• So it is the expected market price of risk in the future that matters

• Since it is a market price, it can vary...
• In CAPM, it is the only risk that should be priced

18 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
12 Amount of risk

• Estimated amount of risk of asset i using CAPM

• Better known as Beta (β)
• It is the (forward looking) covariance of the asset return with the market return,
scaled by the variance of the market return
• βi = COV [rm ,ri ]
V AR[rm ] (Tattoo this somewhere...)
• COV [x, y] = E[(x − E[x])(y − E[y])]
• V AR[x] = COV [x, x] = E[(x − E[x])2] = E[x2] − E[x]2
• It is the future beta we care about. Is it observable?

– No. It doesn’t trade anywhere...

• We can estimate beta from past data

– ri − rf = α
c + β(r
c
m − rf ) + ε

– Dangerous to stop there. Think - does it make sense? Could the future be
different?

19 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 Amount of risk (cont.)

• Many issues:

– Which market portfolio? Roll critique - the market portfolio is unobservable

– Which data set, how far back, what frequency?
– Econometric issues (search “estimated CAPM beta econometric issues”)
– Many potential beta’s...

20 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
13 Worked example

• Risk free rate, Rf = 6%

• Four equally likely states of the world S = [s1, s2, s3, s4]
• Four equally likely market returns, Rm = [−5%, 0%, 5%, 10%]
• Four equally likely firm returns, Ri = [−10%, 0%, 10%, 20%]
• What is the CAPM expected return?
– See “Topic 08 - Manual CAPM beta calculation.xlsx” on Canvas for the
calculations required
– Or you can do it yourself using the definitions of variance and covariance

21 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
14 Estimating the CAPM beta

• In the CAPM, the beta is supposed to be a forward looking estimate of covariance

of risky returns with the market return, scaled by the variance of the market
return
• βi = COV [ri ,rM ]
V AR[rM ]

• (See Excel example file “Topic 08 - Estimating Beta” sheet “OLS” on Canvas
for 7 different ways to calculate beta in Excel.)
• IF we are willing to assume that the future will be like the past, we can use
historical data to estimate beta

– Why should the future be like the past? Has this in fact been the case
generally in the past? No...
– So, if you want to make this assumption, you should have some argument
to support it

22 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 Estimating the CAPM beta (cont.)

– Example arguments:
◦ Regulated industry
◦ Low rate of technological innovation
◦ Single dominant industry leader with strong competitive advantages
◦ Stable economy
– If you know that the future will be different from the past, you should expli-
citly take that into account in your estimate of beta
– Estimating beta from historical data will not be the “right” beta in that case

23 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 Estimating the CAPM beta (cont.)

• To estimate beta, we run the univariate OLS time series regression

– By happy coincidence, the coefficient in a univariate OLS regression (with

an intercept) is given by
COV [ri,t ,rM,t ]
◦ βci = V AR[rM,t ]
◦ The same expression as the theoretical beta!
◦ EXCEPT that theoretical beta is forward looking, while OLS beta depends
on past data
◦ So we can use OLS regression to estimate beta
– ri,t − rtf = α
d +β
i
c f
i (rM,t − rt ) + εi,t

– Univariate = one variable on the right hand side, in this case excess market
return (market return minus risk free rate)
– OLS = Ordinary least squares. Minimises the sum of squared errors
– By convention, we use Greek letters for statistical estimates. Hence alpha
(α) and beta (β) - just “a” and “b” in Greek.

24 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 Estimating the CAPM beta (cont.)

– Note the use of a “hat” (x)

b over the estimates.

◦ This, again, is convention to distinguish between the “true” beta β and

our estimate β.
c

◦ In practice, people get sloppy and you have to infer from the context
whether they mean the true (but unobservable) beta or the OLS estimated
beta
– Note the index variables i and t
◦ i refers to a particular asset
◦ t refers to a particular point in time

25 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 Estimating the CAPM beta (cont.)

– Time-series regression
◦ That means we first select a security i and then
◦ run the regression on returns over time
◦ not to be confused with cross-sectional regressions (regressions across
assets at a fixed point in time)

26 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
15 Arbitrage Pricing Theory (APT)

• Due to Ross 1976

• Assume

– Realised excess returns are driven by a linear N -factor model plus idiosyn-
cratic noise
◦ ri − rf = βi,1λ1 + ... + βi,N λN + εi
– Assume the N-factors and the idiosyncratic noise are all orthogonal (linearly
uncorrelated)
◦ E[λiλj ] = 0, E[λiεj ] = 0 and E[εiεj ] = 0 for all i, j

• Then expected excess return should be a linear combination of the factors

– E[ri] − rf = βi,1λ1 + ... + βi,N λN

27 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
16 CAPM vs APT

• Assume APT with one factor

– E[ri] − rf = βiλ1 thus E[ri] = rf + βiλ
• So, that is the CAPM right?
• Not quite
– λ in the APT does not have to be the market risk premium; in the CAPM
it must be
– In the CAPM, β is the scaled covariance of the asset return with the market
portfolio
– In APT, the β is simply sensitivity to the factor (which might be anything)
• Also, the CAPM is a fully fledged economic model, with agents, expectations,
market clearing equilibrium, etc.
• APT, on the other hand, is more like a framework for a model in which returns
are driven by a set of un-named and assumed uncorrelated risk factors
• So one interpretation of the APT is that is provides a theoretical licence to go
on an empirical fishing trip...
28 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema
 CAPM vs APT (cont.)

• Hundreds of proposed risk factors in the literature!

– Thousands more in lower rated journals and working papers

29 FINANCE 361 Class Notes – University of Auckland – Copyright (C) Dr Paul Geertsema