Investment Theory

COMM 371 - Class 10

Empirical Facts on Stock Returns

“In the 20th century, the United States endured two world wars
and other traumatic and expensive military conflicts; the
Depression; a dozen or so recessions and financial panics; oil
shocks; a flu epidemic; and the resignation of a disgraced
president. Yet the Dow rose from 66 to 11,497.”
Warren Buffett

Prof. Elena Pikulina

October 14, 2015
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

General Overview

So far we studied fixed-income securities. Now we turn to stocks.

I Empirical Facts on Stock Returns (today’s class):

I Introduction, and basic facts on stock returns.
I Portfolio Theory & Passive Investment Strategies:
I How to choose a stock portfolio.
I The Capital Asset Pricing Model (CAPM):
I How the market values stocks.

2 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Today’s Class

I What Stocks Are

I Computing Stock Returns

I Returns as Random Variables

I Stocks vs. Bonds

I Correlation

3 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

What Stocks Are

I Different types of stocks: common stock, non-voting stock, preferred

stock.

I Main features of common stock:

I Voting rights
I Rights to dividends
I Limited liability

I Dividends are the cash flows of common stock. They are:

I discretionary. By contrast, debt payments are mandatory.
I not tax-deductible (because they are not business expenses). By contrast,
debt payments are tax-deductible.
I generally paid quarterly.

4 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Multiple Share Classes of Google Stock

I A-class shares are traded under the symbol GOOGL (originally, GOOG).
I B-class shares have 10-votes-per-share and are owned by insiders only.
I C-class shares began trading on April 4, 2014, under GOOG ticker.

620  
A  class,  GOOGL  
600  
C  class,  GOOG  

580  

560  

540  

520  

500  
3/22/14   5/11/14   6/30/14   8/19/14   10/8/14  
5 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Stock Certificate of Google

6 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Computing Stock Returns

I Return between two dates, 0 and 1 is given by a simple formula:

D1 + P1 − P0
r1 =
P0
I P0 is stock price at date 0, P1 is stock price at date 1 , and D1 is dividend
received at date 1.
I How do we get this formula?

7 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Computing Stock Returns - General Idea

Return between two dates, 0 and 1:

I Invest V0 dollars in the stock at date 0

I If there are any dividend payments between dates 0 and 1, reinvest them
in the stock

I Suppose that value of the stock investment at date 1 is V1 dollars

I Return on the stock is R0,1 is defined by

V0 (1 + R0,1 ) = V1

8 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Computing Stock Returns - in Details

We typically assume that the only dividend payment between dates 0 and 1 is
at date 1

I With V0 dollars, buy x = V0 /P0 shares of the stock at date 0, where P0 is

the date 0 price

I At date 1, these x shares are worth xP1 , where P1 is the date 1 price.
Moreover, they pay a dividend xD1 , where D1 is the date 1 dividend

I Value of the stock investment in period 1 is x(P1 + D1 )

9 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Computing Stock Returns - in Details, Cont-d

I Return on the stock is

x(P1 + D1 ) − V0 x(P1 + D1 ) − xP0 (P1 − P0 ) + D1
R0,1 = = =
V0 xP0 P0

I Return is due to dividend payments and capital gains

I Realized return over the period from t − T to t is

(Pt − Pt−T ) + Dt
Rt−T ,t = ,
Pt−T
where Pt is the price at t, Pt−T is the price at t − T (T > 0), and Dt is
dividend received at t.

10 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Example: Returns on Intel, Inc.

I On August 4, 2015, you spent $144.9 on Intel stock shares (you bought
the shares when the market opened). You sold the shares at closing price
on August 5. Also, the Intel stock paid a dividend of $0.24 per share on
that day. How much money did you earn? What is your return?
I You can look up historical prices at https://ca.finance.yahoo.com.
I On Aug. 04, 2015 the opening price is $28.98 per share.
I On Aug. 05, 2015, the closing price is $29.12 per share.

11 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Returns as Random Variables

12 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Returns as Random Variables

I Future returns are unknown quantities subject to randomness

I From the point of view of today, we view future returns as random

variables

I As random variables, returns are assumed to follow certain distributions

I Distributions could be discrete or continuous

I Frequently, returns are assumed to follow a normal distribution (i.e., they

are assumed to follow a continuous distribution with a bell-curve density)

13 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Parameters

I Important characteristics of distributions are captured by the so-called

parameters. These are theoretical quantities also referred to as population
parameters.

I Examples of parameters include

I the mean (or expected value or expectation) which is a location measure
I the variance (or standard deviation) which is a dispersion measure
I skewness which is a measure of symmetry (or lack thereof)
I and kurtosis which is a measure of heavy-tailedness

I Parameters are computed by taking into account all possible values of the
assumed theoretical distribution.

14 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Expectation, Variance, and Standard Deviation

Let r be a random variable that takes values r1 , . . . , rK with probabilities

p1 , . . . , pK . The expectation, variance, and standard deviation of r as follows:

I The expectation of r is
XK
µr = E (r ) = pk rk
k=1

I The variance of r is
XK
V (r ) = pk (rk − µR )2
k=1

I The standard deviation of r is

p
σ(r ) = V (r )

I For continuous random variables, sums are replaced by integrals.

15 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Estimating Parameters using Statistics based on Data

I Population parameters are unknown quantities since they depend on the
entire distribution which is unknown itself

I However, we can use the available data to obtain proxies of these

parameters. These proxies are called estimators.
I Consider a sample of return series R1 , . . . , RT . Using this sample, we can
I estimate the return (population) expected value by the sample average
PT
R1 + · · · + RT t=1 Rt
R= =
T T
I estimate the return (population) variance by the sample variance
PT  2
2 (R1 − R)2 + · · · + (RT − R)2 t=1 Rt − R
s(R) = =
T −1 T −1
I estimate the return (population) standard deviation by the sample standard
deviation q
s(R) = s(R)2

16 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Stocks vs. Bonds: Risk and Return

Using historical returns 1926-2014, we can compute:

Sample Average Sample Std. Dev.

Large Stocks 12.07% 20.07%
Small Stocks 16.70% 32.13%
Long-Term Govt Bonds 6.11% 9.98%
T-Bills 3.50% 3.11%
Inflation 3.01% 4.10%

Basic empirical fact:

Investments with higher sample standard deviation had higher sample average returns

17 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Stocks vs. Bonds: Cumulative Returns

Stocks, Bonds, Bills, and Inflation: 1926-2014

$100,000.00

$10,000.00
Cumulative Return in Log Scale

$1,000.00

$100.00

$10.00

$1.00

$0.10
1925
1929
1933
1937
1941
1945
1949
1953
1957
1961
1965
1969
1973
1977
1981
1985
1989
1993
1997
2001
2005
2009
2013
Year
Large-Cap Stocks Small-Cap Stocks
Long-Term Government Bonds U.S. Treasury Bills
Inflation
18 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Individual Stocks vs. Indexes

Using monthly returns 1977/01-2015/08, we can compute:

Sample Average Sample St. Dev.

Coca Cola (KO) 15.21 20.54
Disney (DIS) 17.38 28.24
IBM (IBM) 11.31 25.11
Exxon Mobil (XOM) 13.96 17.05
Xerox (XRX) 9.96 36.21
S&P500 8.83 14.95

19 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Diversification Benefits: from Individual Stocks to Broad Indexes

I Note that stocks with higher sample standard deviations tend to have
higher sample average returns

I Note also that the sample standard deviation of the S&P500 is much
smaller than those of the individual stocks. By contrast, the sample
average returns are comparable

I Basic empirical fact:

By holding a portfolio,
we can reduce sample standard deviation
without sacrificing sample average return.
There are benefits to diversification!

20 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Covariance and Correlation

21 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Correlation

I Why diversification works? The answer is related to correlation which is a

measure of how two returns comove in the same or opposite direction.

I Consider two assets with returns R A and R B . The correlation between R A

and R B is defined by:


A B
Cov R A , R B
ρ R ,R = ,
σ (R A ) σ (R B )
where the covariance is defined by

Cov R A , R B = E R A − E R A

  B
R − E RB




.

22 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Properties of Covariance and Correlation

I Covariance and correlation have the same sign. They are:

I positive, if R A and R B tend to increase or decrease at the same time
I zero, if R A and R B are unrelated (uncorrelated)
I negative, if R A tends to be high when R B is low or vice versa.

I Covariance is a linear operator. Let X , Y and Z be random variables and

a, b constants. Then

Cov (X , aY + bZ ) = a Cov (X , Y ) + b Cov (X , Z ).

I The same rule does not apply to correlation!
I Covariance of a random variable with itself is its variance:

Cov (X , X ) = Var (X )

23 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Properties of Correlation

I Correlation is “normalized” covariance.

I Correlation is easier to interpret as it is always a number between -1 and 1.

I Correlation is:
I equal to 1 if there is an exact linear relation with positive slope between X
and Y
(perfectly positively correlated).
I equal to -1 if there is an exact linear relation with negative slope between X
and Y
(perfectly negatively correlated).

24 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

How to Interpret Correlation of Returns?

I If returns of two assets are positively correlated, ρ(rA , rB ) ∈ (0, 1), a

positive return of one asset is more likely to be accompanied by a positive
return by the other asset (compared to the likelihood of a negative
return).

I It is also true that if two assets are positively correlated,

ρ(rA , rB ) ∈ (0, 1), a negative return on one asset is more likely to be
accompanied by a negative return on the other asset (compared to the
likelihood of a positive return).

I Correlation is a probabilistic statement about the extent to which prices

co-move.

I Correlation only measures a linear dependence between asset returns.

25 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Sample Correlation
I Consider two series of returns R A and R B : R1A , . . . , RTA and R1B , . . . , RTB

I The (population) correlation ρ R A , R B is estimated by the sample
correlation
1
PT  A A
 B

T −1 t=1 Rt − R RtB − R
r R A, R B =


s (R A ) s (R B )

1
PT  A
 B

I The numerator T −1 t=1 RtA − R RtB − R is the sample covariance.

I In Excel, sample covariance is computed through the

“COVARIANCE.S(·, ·)” function and sample correlation through the
“CORREL(·, ·)” function.

26 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Correlation of Stock Returns

Using monthly returns 1977/01-2015/08, we can compute the following

correlation matrix:

KO DIS IBM XOM XRX

KO 1
DIS 0.35 1
IBM 0.14 0.35 1
XOM 0.27 0.28 0.30 1
XRX 0.17 0.41 0.50 0.27 1

Note that:

I Stocks are positively correlated

I Correlations can be significant but are still far from perfect

27 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Correlation does not Imply Causation

28 / 29
 Overview
What Stocks Are
Computing Stock Returns
Returns as Random Variables
Stocks vs. Bonds
Correlation

Summary, Suggested Problems and Next Class

I Historically, investments with higher st. deviations tend to have higher

average returns.
I The lower the correlation between stock returns the better diversification
works.

I PPS5 is on Connect
I Marked Assignment 2 is due on Friday, Oct. 16, 5:59pm.

I Next class
I Portfolio Theory

29 / 29