Advanced Lecture 3
Risk and Return
Prof. Evarist Stoja
Calculating returns
• Much of applied finance makes use of returns. The return on an asset between time t-1 and t is the percentage
change in wealth that results from owning the asset
• This change comes from both capital gain, and from cash flows such as dividends or coupons
• The simple return is defined as
For a price For a return index (or adjusted price index)
Pt + Dt - Pt -1 Pt - Pt -1
Rt = Rt =
Pt -1 Pt -1
• Note that because of limited liability, simple returns are bounded from below by –100%, in other words, an
investor cannot lose more than his or her investment in the asset
2
Calculating returns
• Very commonly in finance, we use the continuously compounded (or log) return, defined by
For a price For a return index (or adjusted price index)
rt = ln(1+ Rt )
rt = ln(1+ Rt )
= ln(Pt ) − ln(Pt−1 )
= ln(Pt + Dt ) − ln(Pt−1 )
= ln(Pt / Pt−1 )
• This is for two reasons: firstly, finance theory is often specified in continuous time, and so the concept of
‘return’ in finance often relates to the return over an infinitesimally short period of time; secondly,
continuously compounded returns are more likely to have desirable statistical properties (such as normality)
• The data we have been given (downloaded from Yahoo! Finance) is return index data, and so we do not need
to explicitly allow for dividends in the return computation
3
Calculating returns
• For our two return index series, we can easily
calculate simple returns in Excel as follows
• There is little difference between simple returns
and continuously compounded returns when the
return is small, which is typically the case for short
horizon returns
• For longer horizon returns, or large returns over
short horizon (such as crashes), the difference can
be more substantial, and care must be taken to
ensure that the return calculation is appropriate
• Compute the log returns yourself to check that the
points above do indeed apply for the data we have
4
Statistical functions in Excel
• Security returns are random, and are usually thought of as being drawn from a probability distribution
• Probability distributions are characterised by their moments such as
Mean
Variance - Standard deviation
Skewness
Kurtosis
Covariance - Correlation
• These are useful measures for
summarising financial data and
can be straightforwardly
estimated using Excel functions
• Note that the VARP, STDEVP
and COVAR functions use a
divisor of T, while the VAR and
STDEV functions use a divisor of
T-1
5
Matrix functions in Excel
• A number of applications in finance make use of matrix functions
• These can be easily implemented in Excel using the following functions:
MMULT(A, B) Multiplies the (a x b) matrix, A, by the (b x c) matrix, B, to yield an (a x c) matrix
MINVERSE(C) Computes the inverse of the square matrix C, denoted C-1, which is defined by
C-1C=I, where I is the identity matrix
TRANSPOSE(D) Computes the (b x a) transpose of the (a x b) matrix D, denoted DT
• Like other Excel functions, matrix functions can be nested so complex matrix operations can be easily
performed
• Matrix functions are implemented by selecting the output range with the cursor (so you must first determine
what the dimension of the output matrix is), typing the matrix function with the correct arguments, and then
pressing CTRL+SHIFT+ENTER (as opposed to just ENTER)
• Excel will then perform the matrix calculation and put the output matrix in the area you selected
6
The variance-covariance matrix
• When there is a large number of assets, it is useful to estimate their variance-covariance matrix – this is the
square symmetric matrix that gives the variance of each asset on the diagonal, and the covariance between
each pair of assets on the off-diagonal terms
• Specifically suppose that we have N assets. The variance-covariance matrix is given by
• Although each element of V could be calculated manually (using the VARP and COVAR functions), things
are considerably easier if we use Excel’s matrix functions
7
The variance-covariance matrix
• Suppose that the returns for the N assets over T time periods are in a TxN matrix, denoted R
• Each element of V can be estimated by
8
The covariance matrix
• Using matrix algebra, the covariance matrix can be estimated in a single step as
" r −r … r −r %" r − r … r − r %
1 $
11 1 1T 1
'$ 11 1 N1 N
'
V̂ = $ '$ '
T −1 $ '$ r − r … r − r '
r − r … rNT − rN
$# N1 N '&$# 1T 1 NT N '&
1
= (R − R)T (R − R)
T −1
9
The variance-covariance matrix: example
• For our assets, the variance-covariance matrix can be calculated as follows
• Note how the returns have demeaned using Excel’s matrix functions
10