Risk and Return

Professor Yingzi Zhu

Tsinghua University
 Overview

• Interest rate determinants

• Rates of return for different holding periods
• Risk and risk premiums 风险与风险溢价
• Estimations of return and risk
• Historic returns on risky portfolios
 Real and Nominal Rates of Interest
• The level of interest rates
• the most important macroeconomic factor to consider in one’s
investment analysis
• Nominal interest rate(名义利率）
• A stated interest rate (e.g., 1-year bank deposit rate, 10-year
treasury rate, etc)
• denominated in some unit of account for certain period of time
• Inflation rate
• Price level change: change in consumer’s buying power
• Measured by the percent change in the CPI (the consumer price
index )
• averaging the prices of goods and services in the consumption basket of
an average urban family
• Real interest rate (实际利率)
• the growth rate of consumer’s buying power
Real and Nominal Rates of Interest
• Real Interest Rate (when all rates are known)
 Interest Rate Determinants
• Equilibrium real rates are determined by
supply and demand of funds
– Saver: the supply of funds from Households
– Borrower: the demand for funds from Businesses
– Government can provide either net supply of or
demand for funds
Equilibrium Real Rate of Interest
 Inflation Expectation

• Inflation Expectation
• investors should be concerned with real returns
• When expected inflation rate is higher, nominal interest
rates should be higher
• This higher nominal rate is necessary to maintain the
expected real return offered by an investment
• Interest rate and Inflation Expectation
• Nominal rate should increase one-for-one with
increases in the expected inflation rate
 Interest Rate Determinants: Fisher Effect

• Fisher Effect

– If real rate rates are stable, changes in nominal

rates should predict changes in inflation rates
– difficult to test the hypothesis because the
equilibrium real rate also changes unpredictably
over time
Nominal vs inflation rate
1926-2015
T-Bill Rates, Inflation Rates,
and Real Rates, 1926-2015
 Bills and Inflation, 1926-2015
• Moderate inflation can offset most of the
nominal gains on low-risk investments
• A dollar invested in T-bills from 1926–2012 grew
to $20.25 but with a real value of only $1.55
• the Fisher equation appears to work far better
when inflation is more predictable
• investors can come up with more accurate nominal
interest rate they require to provide an acceptable
real rate of return
中国数据1978-2016
Compare Returns across Holding
Periods
Why Do We Need Annualization?
Example: Zero-Coupon Bond
Annual Percentage Rate (APR)
Effective Annual Rate (EAR)
APR vs EAR
Why Does Compounding Frequency
Matter?
Continuous Compounding
Summary
 Holding Period Return
• Holding period return for an asset paying
dividend
P 1  P 0  D1
HPR 
P0
– HPR = Holding period return
– P0 = Beginning price
– P1 = Ending price
– D1 = Dividend during period one
 Expected Return and Standard Deviation

• There is considerable uncertainty about your

eventual HPR
• Example: suppose you are considering two
alternative investments:
• Treasury bill
• A stock-index fund
• How to quantify the risk of the stock index fund?
 Scenario Returns: Example

State Prob. of State r in State

Excellent .25 0.3100
Good .45 0.1400
Poor .25 -0.0675
Crash .05 -0.5200

We can quantify our beliefs about the state of the market

and the stock-index fund in terms of possible scenarios, with
probabilities
 Scenario Returns: Example
• Expected return

E (r )   p( s )r ( s )
s
• p(s) = Probability of a state
• r(s) = Return if a state occurs
• s = State
E(r) = (.25)(.31) + (.45)(.14) + (.25)(−.0675) + (0.05)(−
0.52)
E(r) = .0976 or 9.76%
 Scenario Returns: Example

• Variance (VAR):

  p s r s   E r 
2
2

s

• Standard Deviation (STD):

STD   2
 Risk premium
• What would it take for you to give up a
guaranteed return and choose something
risky instead?
– 5% more, 10% more, or just a good story?
• Risk aversion
– The degree to which investors are willing to
commit funds to stocks
 Risk Premium
• Excess return: the difference between the
actual rate of return on a risky asset and the
actual risk-free rate
• Risk premium is the expected value of the
excess return
• Risk is measured by standard deviation
 Excess Returns and Risk Premiums
• Scenario analysis of holding-period return of
the stock-index fund
 How to measure Expected Return?
Time Series Analysis of Past Rates of Return
• Estimated Expected Return through historical
data n
1 n
E (r )   p( s)r ( s )   r (s)
s 1 n s 1

– observations in the time series as an equally likely

annual outcome during the sample years and assign it
an equal probability
– If the time series of historical returns fairly represents
the true underlying probability distribution, then the
arithmetic average return from a historical period
provides a forecast of the investment’s expected
future HPR
 How to measure Expected Return?
Time Series Analysis of Past Rates of Return

• The arithmetic average provides an unbiased

estimate of the expected future return.
– But what does the time series tell us about the
actual performance of a portfolio over the past
sample period?
• Geometric (Time-Weighted) Average
g  TV 1/ n
1
TV n  (1  r1 )(1  r2 )...(1  rn )

= Terminal value of the investment

 How to measure Expected Return?
Time series of holding-period returns

Geometric (Time-Weighted) Average

actual performance of a
portfolio over the past sample
period
= Terminal value of the investment
 How to measure Expected Return?
Geometric vs Arithmetic Average
• The larger the swings in rates of return, the
greater the discrepancy between the
arithmetic and geometric averages
• For a normal distribution,
 How to measure risk?
• When thinking about risk, we are interested in
how future outcomes may be deviated (lower)
from the expected return?
• Risk can be captured by variance (or standard
deviation)
– Using historical data with n observations, we
could estimate variance as
 How to measure risk?
Some practical issues
• When to use Geometric mean or Arithmetic
mean?
– historical performance measure: geometric
– Forecast: arithmetic
• Data frequency
– Does it help to improve measurement accuracy?
– Return: no
– Risk: yes
 How to measure risk?
Data Frequency
• Do more frequent observations lead to more
accurate estimates?
• no impact on the accuracy of estimates of
expected return
– the duration of a sample time series improves
accuracy of expected return estimation
• Improves on standard deviation estimation
– because the more frequent observations give us more
information about the distribution of deviations from
the average
– the T-month variance is T times the 1-month variance
 The Reward-to-Volatility:
Sharpe Ratio
• Which investor is smarter: the one who earns
more, or the one who earns more per unit of
stress?
• If we are willing to face risk, then we can
expect to earn a risk premium
– Sharpe ratio: the trade-off between reward and
risk
 Risk Measurements
The Normal Distribution
• Investment management is easier with
normal returns
– Symmetric Returns Standard deviation is a
good measure of risk
– Symmetric Returns  Portfolio returns will be
normal as well
– Only mean and standard deviation needed to
estimate future sceanarios
– Pairwise correlation coefficients summarize the
dependence of returns across securities
 Risk Measurements
The Normal Distribution
• IS normal distribution a good approximation
for asset returns?
 Risk Measurements
Tail risk
• What if excess returns are not normally
distributed?
– STD is no longer a complete measure of risk
• Skewness

• Kurtosis
• Sharpe ratio is not a complete measure of
portfolio performance
 Risk Measurements
Skewed (asymmetric) distributions
• Positive skew – standard deviation over-estimate risk
• Negative skew – standard deviation under-estimate risk
 Risk Measurements
Fat Tailed Distributions
• Kurtosis of a normal distribution Is 0
• Kurtosis greater than 0 (fat tails), standard deviation
will underestimate risk
 Risk Measures： VaR
• Value at Risk (VaR)
– You’ll lose no more than 10% with probability of 5%
– Loss corresponding to a very low percentile of the entire
return distribution, such as the 5- or 1- percentile

VaR(1%)
 Normality and Risk Measures
• Expected Shortfall (ES)
– the expected loss in the worst-case scenario
– More conservative measure of downside risk than VaR
– For normal distribution

Area under
the curve
 Normality and Risk Measures
• Lower Partial Standard Deviation (LPSD)
– Similar to usual standard deviation
– Uses only negative deviations from the risk-free
return
– Addresses the asymmetry in returns
• Sortino Ratio
– The ratio of average excess returns to LPSD
Historic Returns on Risky Portfolios
 Historic Returns on Risky Portfolios
• Normal distribution is generally a good
approximation of returns
– VaR indicates no greater tail risk than equivalent
normal
• Negative skew is present in some portfolio for
some time period
• Positive kurtosis is present in all portfolios for
all the time
Risk-Return: historical perspective
Historical performance for risky assets
 Historic Returns on Risky Portfolios
• The second half of the 20th century offered
the highest average returns
• Firm capitalization is highly skewed to the
right (negative skewness): many small but a
few gigantic firms
 Risk and Return
Average realized returns have generally been
higher for small stocks vs large stocks
 Risk and Return in History
• There is a reward, on average, for investing
in risky assets
• investments with higher risk earn higher
returns

[Link]
[Link]/data_library.html.
 Summary
• Interest rate determinants
• Rates of return for different holding periods
• Risk and risk premiums
• Estimations of return and risk
• Historic returns on risky portfolios