2.

4 Time Series Analysis of Historical Returns

• While historical data analysis can form the basis of expected return
and risk estimation
– Historical data reveals the HPRs realized over specific periods,
– Not what investors expected they would be.
– Therefore, we must use historical data sets to make inferences about
the probability distributions observed HPRs were drawn from.
• If the historical data is representative of the true underlying return dis-
tribution
– This approach will generate sound forecasts.

page 2 of 23
Time Series Analysis of Historical Returns

• The expected return can be estimated by assigning identical proba-

bilities to each historical outcome before calculating the arithmetic
average of the sample HPRs
Xn Xn
E(r) = Pr(s)r(s) = n r(s)
1
s=1 s=1
• Arithmetic average is an unbiased estimate of expected return.
• Geometric (or time-weighted) average return is often preferred when
measuring past performance.
• More on this later.

page 3 of 23
Time Series Analysis of Historical Returns

• We can use our expected return estimates r̂, in generating our risk
estimates σ̂ .
v
u 
√ u n 1 Xn
σ̂ = σ̂ 2 = t × (r(s) − r̂)2
n−1 n
s=1
v
u X
u 1
n
=t (r(s) − r̂)2
n−1
s=1

• By measuring deviations relative to r̂ rather than the true expected

value, we have introduced estimation error, or a degree of freedom
bias.
• We eliminate this by scaling our estimate by n/(n − 1).

page 4 of 23
• The return grow in direct proportion to time.
• Standard deviation grow at the rate of square root of time.
• Suppose the holding period is of N months.
E(r) = √
N E(rm)
σ= N σm
• Example
– 1 month: E(rm) = 1%, σm = 5%. √
– 1 year:E(rm) = 12 × 1% = 12%, σm = 12√× 5% = 17.32%.
– 50 year: E(rm) = 600 × 1% = 600%, σm = 600 × 5% = 122.47%.

page 5 of 23
Time Series Analysis of Historical Returns

• Trade off between

– Return
▷ Measured by the asset’s return in excess of the risk-free rate.
– Risk
▷ Measured by the standard deviation of excess returns.
• Sharpe Ratio
Excess Return
Sharpe Ratio =
Std. Dev. Excess Return
• Widely used in practice when evaluating portfolio performance.

page 6 of 23
Return Distributions

• The process of investment management is dramatically simplified if

asset returns are normally distributed as
– The distribution is completely described by its mean and standard
deviation.
– The risk of the investment is fully described by the standard deviation
of its returns.
– Portfolio comprising stocks with normally distributed returns will
also have normally distributed returns
▷ Meaning the preceding comments will apply equally to such port-
folio.
– The Sharpe ratio is a complete measure of portfolio performance
and thus a tool for investment comparisions.

page 7 of 23
 Normal Distribution
• •Example 1. US
Example: The The US market
market returns
returns (past (past 35 years)
35 years)
70
Frequency
60

50

40

30

20

10

0.12

0.22
0.02
0.04
0.06
0.08

0.14
0.16
0.18

0.24
0.26
0.28
-0.3

-0.2

-0.1
-0.28
-0.26
-0.24
-0.22

-0.18
-0.16
-0.14
-0.12

-0.08
-0.06
-0.04
-0.02
0

0.1

0.2

0.3
Monthly Return

INVESTMENTS | BODIE, KANE, MAR

page 8 of 23
 The Normal Distribution
Normal Distribution
• • Example:
Example 2. Apple
Apple Inc.
Inc. stock stock returns
returns
40
Frequency
35

30

25

20

15

10

0
0.04
0.08
0.12
0.16

0.24
0.28
0.32
0.36

0.44
0.48
0.52
0.56
-0.6

-0.4

-0.2
-0.56
-0.52
-0.48
-0.44

-0.36
-0.32
-0.28
-0.24

-0.16
-0.12
-0.08
-0.04
0

0.2

0.4

0.6
Monthly Return

3 INVESTMENTS | BODIE, KANE, M

page 9 of 23
Normal Distribution
• Mean = 10%.
• SD = 20%.

page 10 of 23
• The proabbility density of a normal distribution
1 (x−µ)2
−
f (x|µ, σ 2) = √ e 2σ2
σ 2π
– Mean: µ.
– SD: σ .
40
Frequency
35

30

25

Normal Distrib
20

15

10

0
0.04
0.08
0.12
0.16

0.24
0.28
0.32
0.36

0.44
0.48
0.52
0.56
-0.6

-0.4

-0.2
-0.56
-0.52
-0.48
-0.44

-0.36
-0.32
-0.28
-0.24

-0.16
-0.12
-0.08
-0.04
0

0.2

0.4

0.6
Monthly Return

page 11 of 23
Return Distributions

• What if excess returns are not normally distributed?

– Standard deviation is no longer a complete measure of risk.
– Sharpe ratio is not a complete measure of portfolio performance.
– Need to consider skewness and kurtosis.

page 12 of 23
Return Distributions

• How do we know it is normal?

• Calculating higher moments of the distribution
– Skewness
▷ Measures the symmetry of the distribution.
▷ Defining R as the asset’s excess return
 
(R − R̄) 3
Skewness = E
σ3

page 13 of 23
Mean = 6%, SD = 17%

page 14 of 23
Positive/Negative Skewness

Negative Skew Positive Skew

page 15 of 23
• Kurtosis
– Measures the fatness of the distribution’s tails relative to normal
 
(R − R̄) 4
Kurtosis = E 4
−3
σ

page 16 of 23
Mean = 10%,SD = 20%

page 17 of 23
Return Distributions

• Implications if return distributions exhibit skewness and/or kurtosis.

– If the return distribution is positively skewed
▷ standard deviation will overestimate risk.
– Conversely, and of even greater concern, if the return distribution is
negatively skewed
▷ standard deviation will underestimate risk.
– When return distributions exhibit ”fat tails”.
▷ standard deviation will underestimate the likelihood of large gains
occurring.
▷ also underestimate the likelihood of large losses occurring.

page 18 of 23
Return Distributions

• A greater likelihood of large negative returns may stem from either

negative skewness or kurtosis.
• Given this, a risk measure would ideally indicate the likelihood an
asset would experience extreme negative returns.
• Three such risk measures that are commonly used in practice are
– Value at Risk, ”VaR”.
– Expected Shortfall, ”ES”.
– Lower Partial Standard Deviation, ”LPSD”.

page 19 of 23
Value at Risk

• VaR is a quantile (q) of a distribution, or the level below which q% of

the distribution’s value lie.
• The 5% VaR, commonly used in practice, is
– The value at the 5th percentile when returns have been sorted in
descending order.
– The value which will be lower than the values of 95% of the distri-
bution, but will be greater than the values of 5% of the distribution.
▷ The ”best of the worst outcomes”.
– Can be derived as Mean − 1.65 × Standard Deviation if returns are
normally distributed.

page 20 of 23
Expected Shortfall

• Expected shortfall
– Tell us the expected loss given that one of the worst case scenarios
eventuates.
– Involves averaging across the lowest 5% of observations.
– Also referred to as the conditional tail expectation.
– More conservative measure of downside risk than VaR.

page 21 of 23
Value at Risk and Expected Shortfall

page 22 of 23
Lower Partial Standard Deviation

• Lower Partial Standard Deviation

– Left-tail standard deviation.
– Focuses solely on negative deviations from the risk-free rate.
▷ Excess returns are evaluated relative to the risk-free rate but not
the mean.
– Is calculated as the square root of total squared negative deviations
from the risk free rate.
▷ It ignores return frequency.
– Is used as the denominator of the Sharpe ratio by practitioners who
use it as a risk measure.
▷ The modified version of the Sharpe ratio called Sortino ratio.

page 23 of 23
 2.3 Risk and Risk Premiums

Expected Return and Standard Deviation

• Given uncertainty about future assets values, investors cannot be cer-

tain about the HPR they will ultimately enjoy.
• We can assign probabilities to possible outcomes and arrive at a weighted
average, or expected return
∑
E(r) = Pr(s)r(s)
s
with
– Pr(s) is the probability of scenario s occurring.
– r(s) is the return given that scenario.

page 2 of 9
• The standard deviation of the rate of return σ is commonly used as a
risk measure.
– Calculated as the square root of variance σ 2, or
√
√ ∑
σ = σ2 = Pr(s)(r(s) − E(r))2
s

page 3 of 9
Example
Consider the following expectations regarding shares in an index fund
over the next year. Each share is currently selling for $100, and a cash
dividend of $4 is paid during the year.
State of the economy Probability Ending price HPR
Boom 0.35 $140 44%
Normal Growth 0.30 $110 14%
Recession 0.35 $80 -16%
Calculate the mean and standard deviation on the HPR.

• Mean
E(r) = 0.35 × 44% + 0.30 × 14% + 0.35 × (−16%) = 14%.

page 4 of 9
• Standard deviation
σ 2 = 0.35 × (0.44 − 0.14)2 + 0.30 × (0.14 − 0.14)2
+ 0.35 × (−0.16 − 0.14)2 = 0.063
σ = 0.251

page 5 of 9
 Expected Return and Standard Deviation
Standard deviation does not differentiate between upside and downside
risk.

• But investors are very concern about the latter.

• If the distribution of possible outcomes is roughly symmetrical around
the mean.
– standard deviation is a reasonable risk proxy.
• If returns are normally distributed.
– Expected return and standard deviation perfectly describe the range
of possible outcomes.
– More on this later.

page 6 of 9
Risk and Risk Premiums

• Risk Premium
– The difference bewteen the expected return and the risk-free rate.
γ = E(r) − rf
• Excess Return
– The difference in the actual return and the actual risk-free rate.
▷ Risk premium is the expected value of the excess return.

page 7 of 9
Concept Check 3
You invest $27000 in corporate bond selling for $900 per $1000 par
value. Over the coming year, the bond will pay interest of $75 per
$1000 of par value. The price of the bond at year’s end will depend
on the level of interest rates that will prevail at that time. You construct
the following scenario analysis:
Interest Rates Probability Year-Ending Bond Price
High 0.2 $850
Unchanged 0.5 $915
Low 0.3 $985
Your alternative investment is a T-bill that yields a sure rate of return of
5%. Calculate the HPR for each scenario, the expected rate of return,
and the risk premium on your investment. What is the expected end-
of-year dollar value of your investment?

page 8 of 9
Solution: Number of bonds bought is 27000/900 = 30.

Interest Rates Prob. Bond Price HPR Value

High 0.2 $850 (75 + 850)/900 - 1 = 0.0278 (75 + 850)30 = 27750
Unchanged 0.5 $915 0.10 29700
Low 0.3 $985 0.1778 31800

• The expected return is 0.2 × 0.0278 + 0.5 × 0.1 + 0.3 × 0.1778 = 0.1089.
• Expected end-of-year dollar value is 29940.
• Risk premium is 0.1089 − 0.05 = 0.0589.

page 9 of 9
 2.2 Comparing Rates of Return for Different Holding Periods

Zero Coupon Bond

• Par = $100.
• Maturity = T
• Price = P
• Total risk free return
100
rf (T ) = −1
P (T )

page 2 of 16
Example: Annualized Rates of Return
Suppose prices of zero-coupon Treasuries with $100 face value and
various maturities are as follows.
Horizon, T Price, P (T ) [100/P (T )] − 1 Total Return
Half-year $97.36 100/97.36-1 = 0.0271 rf (0.5) = 2.71%
1 Year $95.52 100/95.52-1 = 0.0469 rf (1) = 4.69%
25 Year $23.30 100/23.30-1 = 3.2918 rf (25) = 329.18%

page 3 of 16
Holding Period Returns

• The holding period return (HPR) is a measure of the return earned

over a given investment period
Investment Valuet + Investment Incomet − Investment Valuet−1
HPR =
Investment Valuet−1
• The HPR calculation implicitly assumes that
– investment income is received at the end of the holding period.
– to the extent this is incorrect, ignores reinvestment income.

page 4 of 16
Holding Period Returns

• While the calculation below will often yield the same result, this will
not be true when, for example, calculating the HPR on shares which
undergo capitalization changes during the holding period
Pricet + Income (e.g. Dividend)t − Pricet−1
HPR =
Pricet−1
• Given this, you should either use the definition on the preceding slide
or include adjusted prices in the above formula.

page 5 of 16
Summing/Average Returns Over Time

• So far only considered returns over a single period.

• Investors receive returns across multiple time periods.
• Calculating the total returns as the sum of returns over time:

• Arithmetic sum • Geometric sum

∑n ∏
n
rt = rt rt = (1 + rt) − 1
t=1 t=1
• Average return over time:

• Arithmetic mean • Geometric mean

[ n ]1
1∑
n
∏ n
rt =
n
rt rt = (1 + rt) −1
t=1 t=1

page 6 of 16
Example: Returns Over Time
Assume the returns on an asset over 3 consecutive holding periods
are 4%, -6% and 5%, respectively. Given this information
1. What is the arithmetic cumulative return over the entire period?
2. What is the geometric cumulative return over the same period?
3. What is the arithmetic average return over the period?
4. What is the geometric average return over the period?

Solution
1. 3% from
Arithmetic cum. ret. = 4% − 6% + 5% = 3%.
2. 2.648% from
Geometric cum. ret. = (1 + 4%) × (1 − 6%) × (1 + 5%) − 1 = 2.648%.

page 7 of 16
3. 1% from
4% − 6% + 5%
Arithmetic avg. ret. = = 1% .
3
4. 0.875% from
1
Geometric avg. ret. = [(1 + 4%) × (1 − 6%) × (1 + 5%)] −1 = 0.875%.
3

page 8 of 16
Geometric and Arithmetic Returns
• The geometric return is more consistent with the actual return received
by the investor.
• To understand why, assume an investment had the following values
over 3 holding periods.
Year Value Return(%)
1 $1000
2 $2000 100%
3 $1000 -50%
• Arithmetic average return is 25% from
100% − 50%
= 25%.
2
• Geometric average return is 0% from
1
[(1 + 100%) × (1 − 50%)] − 1 = 0%.
2

page 9 of 16
Geometric and Arithmetic Returns
• If the returns on an asset are the same for all periods.
– Geometric average will equal the arithmetic average.
nx̄ n n1
x̄ = = [x̄ ]
n
• If the returns vary across periods.
– the geometric mean will be lower than arithmetic mean.
x+y √
≥ xy
2
this is from
(x − y)2 = (x + y)2 − 4xy ≥ 0
– Higher volatility in returns will exacerbate the difference between
the two measures.

page 10 of 16
Geometric and Arithmetic Returns

• Arithmetic average return

– Good indication of the expected rate of return for an investment
during a future individual year.
• Geometric average return assumes you
– Reinvest all profits back into the stock.
– The reinvested funds earn the rate of return the stock earns in sub-
sequent periods.

page 11 of 16
 Concept Check 4
You invested $1 million at the beginning of 2008 in an S&P 500 stock-
index fund. Given the rate of return for 2008, -40%, what rate of
return in 2009 would have been necessary for your portfolio to recover
to its original value?

Solution Suppose the required rate is r. We have

(1 + r)(1 − 0.40) = 1
so that r = 0.667.

page 12 of 16
Comparing Investments with Different Investment Horizons

• Investments with longer horizons commonly earn greater HPRs.

– HPRs on investments with different holding periods need to be stated
as rates of return for a common period before they are compared.
– Commonly, returns are expressed on an annual basis, or as effective
annual rates (EARs).
– EARs can be calculated using the tools introduced in FINM1001.
1
1 + EAR = [1 + rf (T )] T

▷ A six-month investment with a HPR of 5%

EAR = (1 + 5%)2 − 1 = 0.1025 or 10.25%.

page 13 of 16
Example: Annualized Rates of Return
Suppose prices of zero-coupon Treasuries with $100 face value and
various maturities are as follows.
Horizon, T Price, P (T ) [100/P (T )] − 1 Total Return
Half-year $97.36 100/97.36-1 = 0.0271 rf (0.5) = 2.71%
1 Year $95.52 100/95.52-1 = 0.0469 rf (1) = 4.69%
25 Year $23.30 100/23.30-1 = 3.2918 rf (25) = 329.18%

• EAR = (1 + rf (T ))1/T − 1 = 1.02712 − 1 = 0.0549.

• EAR = (1 + rf (T ))1/T − 1 = 1.04691 − 1 = 0.0469.
• EAR = (1 + rf (T ))1/T − 1 = 3.29181/25 − 1 = 0.06.

page 14 of 16
Discrete v.s. Continuous Compounding

• The EAR is a discrete measure of investment return.

• Continuously compounded rates rcc are sometimes used in practice.
• An EAR can be converted into an rcc as follows
rcc = ln(1 + EAR)
• The difference between the effective and continuously compounded
rates
– Small over short holding periods.
– Large over longer periods.

page 15 of 16
 Concept Check 2
A bank offers two alternative interest schedules for a savings account
of $100000 locked in for 3 years:
a. a monthly rate of 1%.
b. an annually, continuously compounded rate (rcc) of 12%.
Which alternative should you choose?

Solution

a. EAR = (1 + 0.1)12 − 1 = 0.1268.

b. EAR = e12 − 1 = 0.1275.

Choose the continuously compounded rate for its higher EAR.

page 16 of 16
 2.1 Determinants of the Level of Interest Rates.
Determinants of real interest rates include

• Supply side.
– The supply of funds from savers, or those who have more than they
wish to consume today.
• Demand side.
– The demand for funds from those who have less than they wish to
consume or invest today.
• Government’s net demand.
– The government’s net supply of funds or demand for them, both of
which are affected by the actions of the federal reserve(in Australia,
the RBA).

page 2 of 11
The Equilibrium Real Rate of Interest

Why is the supply (demand) curve of real interest rate upward (down-
ward) slopping?

page 3 of 11
 Real and Nominal Rates of Interest
• Nominal interest rate.
– The growth rate of money.
• Real interest rate.
– The growth rate of purchasing power.
• Suppose rn the nominal rate, rr the real rate, and i the inflation rate.
The approximation formula is
rr ≈ rn − i
• The exact relationship
Growth of money
z }| {
1 + rn
|1 +
{zr}r =
1|{z}
+i
Growth of purchasing power
Growth of prices

page 4 of 11
The Theory of Interest
To work out the approximation, we
know
1 + rn = (1 + rr )(1 + i)
= 1 + rr + i + rr i
Since rr i is very small comparing to
rr and i, we have
1 + rn ≈ 1 + rr + i
therefore
rn ≈ rr + i
rr ≈ rn − i

rn = rr + E(i)
page 5 of 11
Example 5.1 Approximating the Real Rate
If the nominal interest rate on a 1-year CD (certificate of deposit) is
8%, and you expect inflation to be 5% over the coming year.
• Using the approximation formula, you expect the real rate of inter-
est to be
rr = 8% − 5% = 3%.
• Using the exact formula, the real rate is
0.08 − 0.05
rr = = 2.86%.
1 + 0.05
• The approximation rule overstates the expected real rate by 0.14%.

page 6 of 11
 Real and Nominal Rates of Interest
Investors are concerned with their real rate of interest given it measures
the increase in their purchasing power. Given this,

• Investors will demand higher nominal rates of return on their invest-

ments as expected inflation increases.
• Fisher (1930) argued that, with stable real interest rate, nominal inter-
est rates should increase one for one with expected inflation.
• This is hard to test (as real rates aren’t constant over time) but nominal
interest rates and inflation move closely together.

page 7 of 11
Interest Rates

page 8 of 11
Interest Rates

page 9 of 11
 Interest Rates
Concept check 1
a. Suppose the real interest rate is 3% per year and the expected in-
flation rate is 8%. What is the nominal interest rate?
b. Suppose the expected inflation rate rises to 10%, but the real rate
is unchanged. What happens to the nominal interest rate?

a. The nominal interest rate is 11.24%, which is from

1 + rn = (1 + rr )(1 + i) = 1.03 × 1.08 = 1.1124.
b. The nominal interest rate is 13.3%, which is from
1 + rn = (1 + rr )(1 + i) = 1.03 × 1.10 = 1.133.

page 10 of 11
Negative Policy Rate

page 11 of 11