CT8 Pu 14

CT8: CMP Upgrade 2013/14 Page 1
Subject CT8
CMP Upgrade 2013/14
CMP Upgrade
This CMP Upgrade lists all significant changes to the Core Reading and the ActEd
material since last year so that you can manually amend your 2013 study material to
make it suitable for study for the 2014 exams. It includes replacement pages and
additional pages where appropriate. Alternatively, you can buy a full replacement set of
up-to-date Course Notes at a significantly reduced price if you have previously bought
the full price Course Notes in this subject. Please see our 2014 Student Brochure for
more details.
This CMP Upgrade contains:
 all changes to the Syllabus objectives and Core Reading
 changes to the ActEd Course Notes, Series X Assignments and Question and
Answer Bank that will make them suitable for study for the 2014 exams.
The Actuarial Education Company © IFE: 2014 Examinations

Page 2 CT8: CMP Upgrade 2013/14
1 Changes to the Syllabus Objectives and Core Reading
1.1 Syllabus Objectives
A new objective has been added as follows:
(i) 9. Discuss the key findings in behavioural finance.
1.2 Core Reading
The Core Reading on the Wilkie model has been greatly reduced in extent. In
particular, all of the Core Reading on pages 23 to 37 inclusive of Chapter 9 of the
Course Notes has been deleted. The revised (and much-shortened) section on the
Wilkie model is included at the end of this upgrade.
A new section of Core Reading on behavioural finance has been added. This Core
Reading, together with the accompanying section from the Course Notes, is included at
the end of this upgrade.
© IFE: 2014 Examinations The Actuarial Education Company

2 Changes to the ActEd Course Notes
Chapter 0
The following list of ST5 exam questions (from April 2005 to April 2013 inclusive) that
refer to behavioural finance and so may now be of relevance to Subject CT8 has been
added:
 April 2005: Q6
 September 2005: Q4
 April 2008: Q6 (ii)
 October 2011: Q8 (ii) and (iii)
 October 2012: Q3.
Chapter 2
A new section has been added containing the new Core Reading on behavioural finance.
This section is included at the end of this upgrade, together with a new exam-style
question that has also been added.
Note that behavioural finance has been combined with stochastic dominance to form a
“new” Chapter 3 in the 2014 Course Notes. Consequently, the chapters numbers from
Chapter 3 (Measures of investment risk) onwards in the 2013 version of the Course
Notes are now one higher in the 2014 version.
Chapter 9
As mentioned above, the section on the Wilkie model has been reduced greatly in
length. In addition, a new exam-style question has been added. Both the revised
section on the Wilkie model and the new question are included at the end of this
upgrade.

Chapter 15
Page 21
The following equation in the middle of the page in terms of St :
St = S 0 exp[(r - 21 s 2 )t + s Z t ]
has been replaced by the following equation in terms of St :
St = S0 exp[(r - q - 21 s 2 )t + s Z t ]
This is the equation that appears in the Core Reading. Both equations are correct.

3 Changes to the Q&A Bank

The following questions, which are all based on the Wilkie model and consequently are
no longer relevant to the syllabus, have been deleted from the Q&A Bank:
 Part 2 : Questions 2.18, 2.19, 2.21 and 2.22
 Part 5: Question 5.2.
In addition, parts (i) and (iii) of Question 2.17 have been edited as follows to reflect the
new Core Reading.
(i) Explain what is meant by the Wilkie model of investment returns. [4]
(iii) State two key differences between the Wilkie model and the continuous-time
lognormal model. [2]
(i) Explain what is meant by the Wilkie model
The Wilkie model attempts to model the processes generating investment returns for
several different types of asset. [½]
It can therefore be used to simulate the possible future development of investment

returns, eg as part of an asset liability modelling exercise. [½]
It is primarily statistically-based – having been estimated using historical data for the
relevant time series involved. [½]
It has a cascade structure. [½]
The key variable is the force of inflation. [½]
This is modelled as a first-order autoregressive process with normally distributed

innovations. [½]
Inflation is assumed to be the driving force behind the other variables, such as: [½]
 share dividends
 share dividend yields
 wage inflation. [½]
It is therefore a particular case of a vector autoregressive moving average (VARMA)

model. [½]
[Maximum 4]

(iii) Two key differences between the models
The lognormal model assumes that investment returns in non-overlapping intervals are
independent and that the expected return does not change over time. In contrast, the
Wilkie model models key investment variables – eg yields – as autoregressive processes
that tend to revert to a long-term mean value. [1]
The Wilkie model models the yields and/or prices and/or returns produced by several
different classes of asset (including equities, conventional and index-linked bonds, and
property), whereas the lognormal model is usually applied only to equity prices. [1]
[Total 2]
The following new questions and solutions have been added to the Q&A Bank.
Question
(i) Within the context of behavioural finance, explain fully what is meant by
overconfidence. [4]
The board of directors of an actively managed investment trust are concerned that the
decisions of the trust’s investment manager may subject to overconfidence bias, which
could adversely affect the performance of the trust.
(ii) Discuss possible actions that the board could take in order to try and limit the
impact of the investment manager’s overconfidence bias. [6]
[Total 10]
Solution
(i) Explain fully overconfidence
People tend to overestimate their own abilities, knowledge and skills. [1]
Moreover, studies show that the discrepancy between accuracy and overconfidence
increases (in all but the simplest tasks) as the respondent is more knowledgeable.
(Accuracy increases to a modest degree but confidence increases to a much larger
degree.) [1]

This may be a result of:
 Hindsight bias – events that happen will be thought of as having been

predictable prior to the event; events that did not happen will be thought of as
having been unlikely ever to happen. [1]
 Confirmation bias – people will tend to look for evidence that confirms their
point of view (and will tend to dismiss evidence that does not justify it). [1]
[Total 4]
(ii) Possible actions to limit the extent of overconfidence bias
The board could require that all investment decisions made by the investment manager
are reviewed by a second investment manager before being implemented. [1]
Alternatively, the management of the investment trust could be split equally between
two investment managers. [1]
Either of these should reduce the impact of overconfidence bias, although the views of
the second investment manager could be subject to similar overconfidence biases as
those of the first manager. [1]
The investment manager could be sent on a training course about behavioural finance,
to make him aware of his possible overconfidence bias. [1]
However, highlighting his possible overconfidence may make him believe that he is no
longer subject to overconfidence bias, making him even more confident in his abilities
and more subject to overconfidence bias. [1]
The board could place tighter constraints on the investment decisions taken by the
investment manager, eg limits could be placed on the size of any transactions and/or on
the size of holdings in individual companies or sectors. [1]
Limiting his actions should limit the scope for biases in his investment decisions but
will also reduce his scope for active investment management and possibly the returns he
achieves. [1]
[Maximum 6]

Question
(i) Define framing and outline how it relates to some of the other behavioural
finance themes. [5]
(ii) The marketing team of a unit trust management company are updating the
promotional literature for their actively managed UK Equity trust. The trust
produced a return of 5% over the last calendar year, compared to 4% for the
FTSE 100 index. Give examples of how the marketing team could present the
past performance of the fund favourably to prospective investors. [4]
(iii) A financial advisor is due to present a range of possible investment strategies to

a high net worth individual. Explain the decisions she will need to make with
regard to the presentation of the possible strategies. [4]
[Total 13]
Solution
(i) Define framing
Framing refers to the fact that the way a choice is presented or “framed” and,
particularly, the wording of a question in terms of gains and losses, can have an
enormous impact on the answer given or the decision made. [1]
Changes in the way a question is framed of only a word or two can have a profound
effect. [½]
In the same way, “structured response” questions are found to convey an implicit range
of acceptable answers. [½]
Framing is related to:
 anchoring and adjustment – as the answer to a question will tend to be anchored

by a suggested answer where one is presented [1]
 prospect theory – as decisions will be influenced by how information is

presented in terms of gains or losses relative to some benchmark [1]
 the effect of options – as the ordering of a range of options can influence the
choice of option or options selected. [1]
[Total 5]

(ii) How to present the past performance of the fund favourably
The marketing team could choose to focus on past performance over a different time
period, during which the unit trust has produced an annualised rate of return that is
higher than 5%, eg 2 years or 5 years. [1]
They could highlight the performance of the unit trust against the FTSE 100, which the
unit trust outperformed by 1% over the last calendar year. [1]
They could highlight the performance of the unit trust against a selection of other
similar actively managed unit trusts that returned less than 5% over the last calendar
year. [1]
They could highlight the performance of the unit trust against a selection of other
similar but passively managed unit trusts that returned less than 5% over the last
calendar year. [1]
They could highlight the performance of the unit trust against other asset classes that
returned less than 5% over the last calendar year, eg cash. [1]
[Maximum 4]
(iii) Decisions with regard to the presentation of the possible investment strategies
She will need to decide:
 how many possible investment strategies to present – remembering that the more
options she presents, the more difficult it may be for the investor to reach a
decision [1]
 which strategies to present – bearing in mind that the investor is more likely to
choose an intermediate option than an outlying one. [1]
In particular, she must decide whether to include the current strategy, and to
highlight it as such, as the investor may tend to stick to that due to status quo
bias / regret aversion. [1]
 how the strategies should be presented – remembering that the choice made may
be influenced by how the options are framed [1]
 the order in which to present the strategies – bearing in mind that the investor
may be more likely to choose the first strategy (primary effect) or the last
strategy (the recency effect). [1]
[Maximum 4]

Question
A zero-coupon bond (ZCB) promises to pay a return of 8% when it matures in a year’s

time, assuming that it doesn’t default. In the event of default, which has a probability of
10%, it will yield a return of either 6% or 0%, with equal likelihood.
A 1-year risk-free government ZCB provides a return of 5%.
(i) Calculate the following measures of investment risk for the risky ZCB:
(a) downside semi-variance
(b) shortfall probability relative to the risk-free rate
(c) the expected shortfall relative to the risk-free rate conditional on a
shortfall occurring. [4]
A bond analyst is going to use a two-state intensity-based model to derive risk-neutral

transition intensities.
(ii) (a) State the general risk-neutral pricing formula for a ZCB subject to default
risk in a two-state intensity-based model with a deterministic transition
intensity. Define all notation used.
(b) Starting from the formula in (ii)(a), derive the corresponding formula for
the risk-neutral transition intensity function in terms of the ZCB price. [6]
[Total 10]
Solution
(i)(a) Downside semi-variance
The expected return on the risky ZCB is equal to:
0.9  8%  0.05  6%  0.05  0%  7.5% [1]
Consequently, the downside semi-variance is given by:
 7.5  6 2  0.05   7.5  0 2  0.05  2.925%% [1]
(i)(b) Shortfall probability relative to risk-free rate
The only outcome that provides a lower return than the risk-free rate of 5% is when the
ZCB defaults and returns 0%. The shortfall probability is therefore 5%. [1]

(i)(c) Expected shortfall conditional on a shortfall occurring
This is equal to the unconditional expected shortfall divided by the shortfall probability.
Again, there is only one outcome that is worse than the risk-free rate of 5% and so this
is equal to:
 5  0   0.05
 5% [1]
0.05
[Total 4]
(ii)(a) General risk-neutral pricing formula for a ZCB
È Ï Ê T ˆ ¸˘
B (t , T ) = e
- r (T -t ) Í1 - (1 - d ) ÔÌ1 - exp Á - l ( s )ds ˜ Ô˝˙
Í Ë t
Ú ¯ Ô˛˙˚
[1]
Î ÔÓ
where:
 t and T are the current time and the maturity date of the ZCB
 r is the constant risk-free force of interest
  is the assumed (constant) recovery rate
  ( s) is the risk-neutral transition intensity. [1]
(ii)(b) Derive formula for risk-neutral transition intensity function
Multiplying both sides of the ZCB pricing formula by e r (T -t ) and then multiplying out
the right-hand side, gives:
   
e r (T t ) B(t , T )  1  1  exp     ( s )ds     1  exp     ( s )ds  
T T
  t    t 
 exp     ( s )ds      exp     ( s )ds 

T T
 t   t 
 (1   ) exp     ( s )ds   
T
[1]
 t 

Moving the constant d to the left-hand side, then taking logs, we get:
e r (T -t ) B (t , T ) - d = (1 - d ) exp Ê - Ú l ( s )ds ˆ
T
Ë t ¯
T
 log ÈÎ er (T -t ) B(t , T ) - d ˘˚ = log(1 - d ) - Ú l ( s)ds [1]
t
Differentiating both sides with respect to T then gives:
∂
log ÈÎ er (T -t ) B(t , T ) - d ˘˚ = - l (T ) [1]
∂T
Remember that differentiating an integral (with respect to the upper limit) takes you
back to the original function.
Finally, renaming the variable T as s , swapping the two sides and flipping the signs
gives:
∂
l ( s ) = - log ÈÍ e ( ) B (t , s ) - d ˘˙
r s -t
[1]
∂s Î ˚
[Total 6]
Question
(i) Explain what is meant by a short rate model of interest rates and how such
models can be used to price interest rate derivatives. [3]
(ii) Explain what is meant by a one-factor short rate model of interest rates. [2]
[Total 5]
Solution
(i) Short rate model of interest rates
Short rate models are used to model the term structure of interest rates as a function of
the short rate r (t ) . [½]
In theory, this is the interest rate that applies over the next instant of time, of length dt .
[½]
In practice, the short rate is an overnight rate, ie the force of interest earned when
money is lent today and received back with interest the following day. [½]

r(t) itself is usually assumed to be a diffusion or Ito process, with a stochastic

differential equation of the form:
dr (t )  a (t , r (t )) dt  b(t , r (t )) dWt
where Wt is a standard Brownian motion under the risk-neutral probability measure, Q.
[1]
The price at time t of an interest rate derivative that pays X T at maturity date T  t can
then be found from:
È ˘
Vt = EQ Í exp Ê - Ú r (u )du ˆ X T Ft ˙
T
[1]
Î Ë t ¯ ˚
[Maximum 3]
(ii) Explain what is meant by a one-factor short rate model
A one-factor short rate model is one in which the short rate, and hence the term
structure as a whole, is assumed to be influenced by a single source of randomness. [½]
The prices of all bonds (of all maturities) and interest-rate derivatives must therefore
move together. [½]
The single source of randomness is typically assumed to be a standard Brownian motion

process. [½]
The short rate is therefore usually modelled as an Ito process. [½]

[Total 2]

Question
An exotic forward provides a payoff equal to the square root of the share price at
maturity time T less the square root of the delivery price, K.
(i) (a) Assuming that the Black-Scholes assumptions apply, use risk-neutral
pricing to derive a formula for the price at time t of the forward on a
non-dividend-paying share.
(b) Derive the corresponding formula for the vega of the forward. [6]
(ii) (a) Explain why an investor might want to vega hedge their portfolio.
(b) Use the result that St  (d1 )  Ke r (T t ) (d 2 )  0 , where d1 and d 2 are
defined as on page 47 in the Tables, to show that the formula for the vega
of a European call option is  call  St (d1 ) T  t . [6]
The current price of the share is $1, which is also the delivery price of the forward. The
risk-free force of interest is 5%, the volatility of the underlying share, which pays no
dividends, is 20% and the forward has one year to delivery.
(iii) An investor has a long position in 1,000 exotic forwards. Find the vega-hedged
portfolio for this position involving standard European call options on the
underlying share and also the underlying share itself. [8]
[Total 20]

Solution
(i)(a) Derive formula for price of forward
The general risk-neutral formula for pricing a derivative at time t < T is:
Vt = e - r (T -t ) EQ ÈÎ X T Ft ˘˚ [½]
In this instance, the payoff function of the derivative is:
X T = ST½ - K ½
So, the price will be given by:
Vt = e - r (T -t ) EQ ÈÎ ST½ - K ½ Ft ˘˚ [1]
Black-Scholes assumes that under the risk-neutral measure Q , the underlying share
price follows geometric Brownian motion with drift r and volatility s and so:
Î ( )
log ST ~ N È log St + r - ½s 2 (T - t ), s 2 (T - t ) ˘
˚
[½]
Hence, using the formula for the moments of a lognormal distribution from page 14 in
the Tables, with r = ½ , the price will be given by:
È ½(log St +( r -½s 2 )(T -t )) + ½ ¥½ 2 s 2 (T -t ) ˘

Vt = e - r (T -t ) Í e - K½ ˙ [1]
Î ˚
Note that the formula in the Tables also works for non-integer values of r.
This simplifies to:
Vt = St½ e
( )
-½ r + ¼s 2 (T -t )
- K ½ e - r (T -t ) [1]
[Total 4]

(i)(b) Formula for the vega of the forward
The vega of a derivative with price f based on an underlying share with volatility s is
defined as:
∂f
n = [½]
∂s
So, here:
n = -¼s (T - t ) St½ e
( )
-½ r + ¼s 2 (T -t )
[1½]
[Total 2]
(ii)(a) Why vega hedge a portfolio?
A vega-hedged portfolio is one whose overall vega, which is equal to the sum of the
vegas of the constituent securities, is close to zero. [½]
Consequently, the value of such a portfolio will be relatively insensitive to changes in

the volatility of the underlying share. [½]
An investor might therefore wish to vega hedge in order to:
 protect the value of a portfolio against (small) changes in the volatility of the
underlying share. [½]
 compensate for the fact that the volatility of the share is unknown, as it cannot
be observed directly. It is less important to have an accurate estimate of the
volatility if vega is low and hence has little effect on the portfolio’s value. [½]
[Total 2]
(ii)(b) Show that vega of European call option is  call  St  (d1 ) T  t
Differentiating the Black-Scholes formula for a European call option with respect to s
using the product and chain rules for differentiation gives:
∂d1 ∂d
n call = Stf (d1 ) - Ke - r (T -t )f (d 2 ) 2
∂s ∂s
where f ( x) = 1
2p ( )
exp - 12 x 2 is the probability density function of the standard normal
distribution (from page 11 of the Tables). [1]

Now:
d 2  d1   T  t
Thus:
d 2 d1
  T t [½]
 
Hence:
d1  d 
 call  St  (d1 )  Ke  r (T t ) (d 2 )   1  T  t 
   
d1 
 St  (d1 )  Ke  r (T t ) (d 2 )   Ke  r (T t ) (d 2 ) T  t [1]
  
So, using the result that:
St (d1 )  Ke r (T t ) (d 2 )  0
we have:
d1
 call   0  Ke r (T t ) (d 2 ) T  t

 Ke  r (T t ) (d 2 ) T  t [1]
Or equally, given that:
St (d1 )  Ke r (T t ) (d 2 )
this can be written as:
 call  St (d1 ) T  t [½]

[Total 4]

(iii) Find vega-hedged portfolio
Using the information given in the question, together with the formulae found earlier in
the question, the price and vega of the forward are:
Vt = St½ e
( )
-½ r + ¼s 2 (T -t )
- K ½ e - r (T -t )
= 1¥ e
(
-½ 0.05 + ¼ ¥0.22 ¥1 ) - 1 ¥ e -0.05¥1
= 0.019216 [1]
n = -¼s (T - t ) St½ e
(
-½ r + ¼s 2 (T -t ))
= -¼ ¥ 0.2 ¥ 1 ¥ 1 ¥ e
(
-½ 0.05 + ¼ ¥0.22 ¥1 )
= -0.048522 [1]
Using the Black-Scholes formula on page 47 in the Tables:
d1 

log 1  0.05  ½  0.22  1   0.35 [½]
0.2  1
d 2  0.35  0.2  1  0.15 [½]
So, the price of the European call option is:
ct  1   (0.35)  1  e 0.051 (0.15)
 0.63683  0.951229  0.55962
 0.104503 [1]
and its vega is equal to:
 call  St  (d1 ) T  t
1 2
 1 e½0.35  1
2
 0.375240 [1]
Finally, the vega of a share is equal to zero. [½]

So, to vega hedge the long position in 1,000 forwards, we need to find the number of
European calls x and shares y such that:
1, 000  0.019216  0.104503 x  1  y [½]
1, 000  0.048522  0.375240 x  0  y [½]
Solving these equations gives:
x  129.31 [½]
y  32.729 [½]
ie we need to short sell 129.31 calls and buy 32.729 shares. [½]
[Total 8]
This portfolio will respond in a similar way to the exotic forward to (small) changes in
the volatility of the underlying share. Strictly speaking, in order to vega hedge the
exotic forward we need to take the opposite positions to those found above, ie buy
129.31 calls and short sell 32.729 shares.

4 Changes to the X assignments
Assignment X1
A new question on behavioural finance has been added. The question and solution are
as follows.
Question
(i) State the principal theme of behavioural finance. State the assumption of
expected utility theory that is challenged by this theme. [2]
(ii) Outline what is meant by Prospect Theory. [2]
(iii) Briefly describe the behavioural finance theme categorising the behaviour of the
people in each of the following cases.
(a) Short-term interest rates have remained at historically low levels for
several years now and there is no compelling reason for them to change.
However, Saver A continues to expect them to increase anytime soon.
(b) In 2011, Investor B invests his spare cash in Share X, which produces a
return of 5%. In 2012, he splits his money equally between Shares Y and
Z, which return +15% and –5% respectively. He is less happy with the
performance of his share portfolio in 2012 than in 2011. [3]
[Total 7]
Solution
(i) Principal theme of behavioural finance
The field of behavioural finance looks at how a variety of mental biases and decision-
making errors affect financial decisions. It relates to the psychology that underlies and
drives financial decision-making behaviour. [1]
Behavioural finance challenges the assumption of expected utility theory that decisions
are made on the basis of maximising the expected value of utility under the investor’s
particular beliefs about the probability of different outcomes. [1]
[Total 2]

(ii) Prospect theory
This is a theory of how individuals make decisions when faced with risk and
uncertainty. It replaces the conventional risk-averse / decreasing marginal utility theory
based on total wealth with a concept of value defined in terms of gains and losses
relative to a reference point. [1]
Individuals are assumed to be risk-averse when facing gains relative to the reference
point, but risk-seeking when facing losses relative to the reference point. [½]
This generates utility curves with a point of inflexion at the chosen reference point. [½]
[Total 2]
(iii) Describe the behavioural finance theme
(a) This is an example of anchoring and adjustment. [½]
The saver’s expectations with regard to future short-term interest rates are
anchored by past experience of higher interest rates. Consequently, he expects
interest rates to rise back to average historical levels even when there is no
evidence that they are likely to do so. [1]
(b) This is an example of mental accounting. [½]
The investor has achieved a return of 5% on his share investment in each year
and so should be equally happy. However, he places gains and losses into
different mental accounts with different weightings and so the unhappiness
resulting from the 5% loss on Share Z outweighs the additional happiness from
the 15% return on Share Y. [1]
[Total 3]

Assignment X2
Two new questions have been added on the capital asset pricing model (CAPM), and
arbitrage pricing theory (APT) respectively. The questions and solutions are as follows.
Question
(i) Explain what the separation theorem implies about optimal investment
strategies. [2]
(ii) Explain why an individual investor wouldn’t hold the market portfolio as part of
her investment portfolio in practice. [2]
You are given the following historical information for a share in Company ABC and for
a portfolio of 100 shares.
Return (% pa) Standard deviation beta

of return (% pa)
ABC 8.5 20 0.7
Portfolio 10.5 16 1.1
(iii) Use these results to derive the expected return on the market portfolio and the
risk-free rate of return assuming the CAPM applies. [3]
A student has commented that ABC’s lower return and higher standard deviation,
relative to the 100-share portfolio, contradicts the predictions of the CAPM.
(iv) Discuss the student’s comment. [3]

[Total 10]
Solution
(i) What the separation theorem implies about optimal investment strategies
The separation theorem states that, under the assumptions of CAPM, the optimal
combination of risky assets for an investor can be determined without any knowledge of
their preferences towards risk and return. [1]
This combination is the market portfolio, which consists of all risky assets held in
proportion to their market capitalisation. [1]
[Total 2]

(ii) Why an investor wouldn’t hold the market portfolio in practice
In practice, the investor won’t hold the market portfolio because:
 It is impossible to identify exactly the constituents of the market portfolio, as in

principle it contains all risky assets including shares, property, bonds,
commodities and human capital, many of which are never traded and have no
market values. [1]
 Even if it were possible to identify exactly the constituents of the market

portfolio, many of the different risky assets required (eg human capital) cannot
be invested in, in practice. [1]
[Total 2]
(iii) Derive expected return on market portfolio and risk-free rate of return
According to the CAPM security market line, the expected return on any portfolio,
efficient or otherwise, including a single risky stock is given by:
E p = r + b p ( Em - r )
where:
 r is the risk-free rate of return
 Em is the expected return on the market portfolio
 bp =
(
Cov R p , Rm ) [1]
Var ( Rm )
For ABC shares we have:
(1) 8.5 = r + 0.7 ¥ ( Em - r ) = 0.3r + 0.7 Em [½]
whereas for the 100-share portfolio:
(2) 10.5 = r + 1.1 ¥ ( Em - r ) = -0.1r + 1.1Em [½]
(1) + 3×(2) gives:
40 = 4 ¥ Em ,  Em = 10% [½]

Substituting this back into (1) gives:
8.5 = 0.3r + 0.7 ¥ 10 ,  r = 5% [½]

[Total 3]
(iv) Discuss the student’s comment
The student’s reasoning is flawed because according to CAPM, what counts in the
determination of expected returns for inefficient portfolios such as we have here is b
and not the standard deviation. [1]
The student believes that the higher standard deviation of ABC (not surprisingly higher
than the standard deviation of the well-diversified, 100-stock portfolio) justifies a
higher return for ABC. [1]
In fact, ABC has a lower b than that of the 100-share portfolio and so, r and Em
being the same for both, the return on ABC should be lower. [1]
[Total 3]
Question
(i) Outline the main difficulties with testing the arbitrage pricing theory (APT) in
practice. [3]
A modeller is developing a 2-factor APT model and has estimated the following
parameter values in respect of three securities:
Expected return Exposure to factor 1, Exposure to factor 2

Securit
% bi ,1 bi ,2
y
1 8 5 7.5
2 12 12.5 10
3 15 12.5 15
(ii) Use the data in the table to estimate numerical values for l0 , l1 and l2 , and
explain what your answers represent. [5]
[Total 8]

Solution
(i) Main difficulties with testing the arbitrage pricing theory (APT) in practice
The theory does not specify the number of factors that influence returns, nor what the
individual factors are. [1]
This means that any test of the APT is a joint test of both the particular specification
used and the theory in general. In particular, the failure of the model used to predict
returns accurately could be due to either an error in its specification or to a failing of the
theory itself (or both). [1]
It is difficult to estimate numerical values for both the exposure to each factor for each
security, ie the “b’s” and the risk premia, ie the “ l ’s” and any estimate obtained will of
course be subject to random variation. [1]
[Total 3]
(ii) Estimate numerical values for λ0 , λ1 and λ2
Recall that in the APT with just two factors (as we have here), the expected return Ei
on Security i is given by:
Ei = l0 + l1bi ,1 + l2 bi ,2
So, using the data in the table we have the following equations for the three securities:
(1) 8 = l0 + 5l1 + 7.5l2
(2) 12 = l0 + 12.5l1 + 10l2
(3) 15 = l0 + 12.5l1 + 15l2 [1]
Deducting (2) from (3) gives:
3 = 5l2
ie l2 = 0.6 [1]

Substituting this back into (1), (2) and (3) gives:
(4) 3.5 = l0 + 5l1
(5) 6 = l0 + 12.5l1
(6) 6 = l0 + 12.5l1
(5) – (4) then gives:
2.5 = 7.5l1
1
ie l1 = = 0.333 [1]
3
Finally, substituting this value for l1 into (4) above gives:
1
3.5 = l0 + 5 ¥
3
11
ie l0 = = 1.833 [1]
6
l0 = 1.833 represents the element of expected return that is independent of the risky
factors 1 and 2. [½]
l1 = 0.333 and l2 = 0.6 represent the additional expected return stemming from a unit
increase in bi ,1 and bi ,2 respectively. [½]
[Total 5]

Question X2.5
A new part has been added to this question. The new question part and solution are as
follows.
Question
(iii) A modeller uses the lognormal model to run 10,000 projections of a share price
over the next 120 months. Explain whether the longitudinal and cross-sectional
properties will be the same or different for the following two data sets:
(a) next month’s share price given this month’s share price
(b) the log returns over the next month. [4]
Solution
(iii) Longitudinal and cross-sectional properties
(a) Next month’s share price given this month’s share price
The longitudinal and cross-sectional properties will differ due to the non-stationarity of
the lognormal model. [½]
Provided the model has a positive drift parameter m , then the share price will drift
upwards geometrically without limit over time. This means, for example, that the
average projected share price over the next 120 months from any particular simulation
is likely to be much greater than the average value over 10,000 projections of Month 1’s
share price given the current share price at time 0. Consequently, the mean value of the
first and longitudinal sample of observations for next month’s share price given this
month’s share price, will be greater than that of the second, cross-sectional sample.
[1½]
[Total 2]
(b) The log returns over the next month
The longitudinal and cross-sectional properties will be the same for the log returns over
the next month. [½]
This is because the log returns over the next month are assumed to be both stationary
and to be independent of past history including the current share price. [1]

Consequently, if the cross-sectional and longitudinal samples are sufficiently large,

ie the number of projections is large and each projection is sufficiently long, then the
two samples should have very similar statistical properties. [½]
[Total 2]
Questions X2.7 to X2.9
These questions on the Wilkie model have been deleted as they are no longer relevant to
the syllabus.
Assignment X4
Question X4.3
This question, which is still relevant to this syllabus, has been deleted and replaced by
the following question on risk-neutral pricing.
Question
An investment bank has developed a new exotic derivative, which will pay an amount
equal to the share price at maturity multiplied by the share price at maturity less one
dollar. Let T be the maturity date of the derivative and r be the risk-free force of
interest and assume that the Black-Scholes analysis applies.
(i) Use risk-neutral valuation to derive the pricing formula for this derivative at
time t < T , based on a share that pays no dividends. [7]
(ii) (a) Derive the corresponding formula for the delta of the derivative.
(b) Derive a condition for the range of values for the current share price for
which delta is positive and comment on what your answer suggests for
derivatives of this type with differing terms.
(c) Derive the corresponding formula for the gamma of the derivative and
comment on the sign of gamma. [4]
[Total 11]

Solution
(i) Risk-neutral pricing formula for the derivative
The general risk-neutral formula for pricing a derivative at time t < T is:
Vt = e - r (T -t ) EQ ÈÎ X T Ft ˘˚ [½]
In this instance, the payoff function of the derivative is:
X T = ST ( ST - 1) = ST2 - ST
So, the price will be given by:
Vt = e - r (T -t ) EQ ÈÎ ST2 - ST Ft ˘˚ [1]
Given that the share price at maturity can take any value from zero upwards, this can be
evaluated as:
È• • ˘
- r (T -t )
(1) Vt = e Í Ú ST f ( ST St ) dST - Ú ST f ( ST St ) dST ˙
2
[1]
ÍÎ 0 0 ˙˚
where:
 f ( ST St ) is the probability density function for ST given St
 The assumption of independent increments underlying Brownian motion means

we don’t need to condition on the full past history of the share price.
Under the risk-neutral measure Q , the share price follows geometric Brownian motion
with drift r and so:
Î ( )
log ST ~ N È log St + r - ½s 2 (T - t ), s 2 (T - t ) ˘
˚
[½]

So, using the formula for the truncated moments of a lognormal distribution on page 18
in the Tables, (with k = 2 for the first integral and k = 1 for the second integral) (1)
above becomes:
Í
- r (T -t ) Í
( )
È e 2(log St + ( r -½s 2 )(T -t )) + ½ ¥ 4s 2 (T -t ) F(U ) - F( L ) ˘
{ 2 2 }˙
˙
Vt = e
Í
Í- e t
Î ( 2 2
log S + ( r -½s )(T -t ) + ½s (T -t )
)
{F(U1 ) - F( L1 )} ˙
˙
˚
Î (
= e - r (T -t ) ÈÍ St2 e(2 r +s
2
)(T -t )
){F(U ) - F(L )} - S (e ){F(U ) - F(L )}˘˙˚
2 2 t
r (T -t )
1 1
(
= St2 e( r +s
2
)(T -t )
){F(U ) - F(L )} - S {F(U ) - F(L )}
2 2 t 1 1
[1½]
Now, also from page 18 in the Tables:
log(•) - log St - (r - ½s 2 )(T - t )

 U2 = - 2s T - t = • [¼]
s T -t
log(0) - log St - (r - ½s 2 )(T - t )

 L2 = - 2s T - t = -• [¼]
s T -t
log(•) - log St - (r - ½s 2 )(T - t )

 U1 = -s T - t = • [¼]
s T -t
log(0) - log St - (r - ½s 2 )(T - t )

 L1 = - s T - t = -• [¼]
s T -t
So:
 F(U 2 ) = 1
 F( L2 ) = 0
 F(U1 ) = 1
 F( L1 ) = 0 [½]

Hence:
(
Vt = St2 e( r +s
2
)(T -t )
) {1 - 0} - S {1 - 0}
t
2
= St2 e( r +s )(T -t )
- St [1]
[Total 7]
Alternatively, as the lognormal distribution isn’t actually truncated here, you could
instead use the formula for the (untruncated) moments of the lognormal distribution on
page 14 in the Tables, with r = 2 for the first integral and r = 1 for the second integral.
So:
È• • ˘
- r (T -t )
Vt = e Í Ú ST f ( ST St ) dST - Ú ST f ( ST St ) dST ˙
2
ÍÎ 0 0 ˙˚
Í
= e - r (T -t ) Í
(
È e2(log St + ( r -½s 2 )(T -t )) + ½ ¥ 4s 2 (T -t )
)˘˙˙ [2]
( 2 2
Í - elog St + ( r -½s )(T -t ) + ½s (T -t )
ÍÎ ) ˙
˙˚
= e - r (T -t ) È St2 e(2 r +s - St e r (T -t ) ˘
2
)(T -t )
ÍÎ ˙˚
2
= St2 e ( r +s )(T -t )
- St [2]
(ii)(a) Formula for the delta of the derivative
Recall that for a derivative with price ft delta is defined as:
∂f t
D=
∂St
So, here:
2
D = 2 St e( r +s )(T -t )
-1 [1]

(ii)(b) Range of values for the current share price for which delta is positive
Delta will be positive if:
2
2 St e( r +s )(T -t )
-1 > 0
2
St > ½ e - ( r +s )(T -t )
[½]
As the term on the right-hand side decreases as T increases, this indicates that delta is
more likely to be positive for a derivative of this type with a longer outstanding term.[½]
(ii)(c) Derive formula for the gamma of the derivative
Recall that:
∂D
G=
∂St
So for this derivative:
2
G = 2e( r +s )(T -t )
[1]
So, gamma is always positive. [½]
This reflects the fact that the value of the derivative depends directly upon the square of
the share price and hence is a convex function of the share price. [½]
[Total 4]

5 Other tuition services

In addition to this CMP Upgrade you might find the following helpful with your study.
5.1 Study material
We offer the following study material in Subject CT8:

 Mock Exam A
 Additional Mock Pack
 ASET (ActEd Solutions with Exam Technique) and Mini-ASET
 Sound Revision
 Revision Notes
 Flashcards.
For further details on ActEd’s study materials, please refer to the 2014 Student
Brochure, which is available from the ActEd website at www.ActEd.co.uk.
5.2 Tutorials
We offer the following tutorials in Subject CT8:
 a Taught Course (lasting four days)
 a set of Regular Tutorials (usually lasting two or three full days)
 a Block Tutorial (lasting two or three full days)
 a Revision Day (lasting one full day)
 an Online Classroom – a comprehensive collection of over 100 tutorial units,

including both teaching units and guided questions. You can watch a sample of
the Online Classroom tutorial units on the ActEd website.
For further details on ActEd’s tutorials, please refer to our latest Tuition Bulletin, which
is available from the ActEd website at www.ActEd.co.uk.

5.3 Marking
You can have your attempts at any of our assignments or mock exams marked by
ActEd. When marking your scripts, we aim to provide specific advice to improve your
chances of success in the exam and to return your scripts as quickly as possible.
For further details on ActEd’s marking services, please refer to the 2014 Student
Brochure, which is available from the ActEd website at www.ActEd.co.uk.

6 Feedback on the study material

ActEd is always pleased to get feedback from students about any aspect of our study
programmes. Please let us know if you have any specific comments (eg about certain
sections of the notes or particular questions) or general suggestions about how we can
improve the study material. We will incorporate as many of your suggestions as we can
when we update the course material each year.
If you have any comments on this course please send them by email to CT8@bpp.com
or by fax to 01235 550085.

All study material produced by ActEd is copyright and is sold
for the exclusive use of the purchaser. The copyright is owned
by Institute and Faculty Education Limited, a subsidiary of
the Institute and Faculty of Actuaries.
Unless prior authority is granted by ActEd, you may not hire

out, lend, give out, sell, store or transmit electronically or
photocopy any part of the study material.
You must take care of your study material to ensure that it is

not used or copied by anybody else.
Legal action will be taken if these terms are infringed. In

addition, we may seek to take disciplinary action through the
profession or through your employer.
These conditions remain in force after you have finished using

the course.

CT8-03: Stochastic dominance and behavioural finance Page 11
3 Behavioural finance
3.1 Introduction
The field of behavioural finance looks at how a variety of mental biases and
decision-making errors affect financial decisions. It relates to the psychology that
underlies and drives financial decision-making behaviour.
Although traditional economic theory assumes that investors always act rationally,
ie with the aim of maximising expected utility, experimental and actual evidence
suggests that this may not always be entirely the case.
Much of the work to date has concentrated on the impact on prices in capital
markets (indeed, some “contrarian” investment funds are run on the basis of
taking advantage of errors made by other investors).
A contrarian fund is one that tends to take the opposite view to the rest of the market –
ie it will tend to buy shares when most people in the market are recommending sell and
vice versa. For example, it might sell shares when the market is “high” or rising on the
basis that the market tends to overreact to positive news and so is likely to be
overvalued.
In investment consultancy, behavioural arguments can be applied to trustees and

used to justify proposed investment management structures.
In practice, the management of an investment portfolio in accordance with an

investment objective will require both:
 an investment strategy
 a choice of investment management structure.
Once these have been determined, the implementation of the chosen investment strategy
will then require the selection of the individual investments and investment managers.
If it is recognised that the trustees responsible for the direction of investment policy are
subject to the types of mental bias identified by behavioural finance, then the
recommended investment management structure may be chosen to reflect those biases.
Likewise, if capital markets are indeed influenced by behavioural factors, then those
investors who recognise this may be able to exploit this knowledge to the disadvantage
of investors who don’t.

Page 12 CT8-03: Stochastic dominance and behavioural finance
Common themes found in research on behavioural finance include:

 anchoring and adjustment
 prospect theory
 framing (and question wording)
 myopic loss aversion
 estimating probabilities
 overconfidence
 mental accounting
 the effect of options.
3.2 Anchoring and adjustment
Anchoring is a term used to explain how people will produce estimates. They start
with an initial idea of the answer (“the anchor”). They then adjust away from this
initial anchor to arrive at their final judgement.
Thus, people base perceptions on past experience or “expert” opinion, which they
amend to allow for evident differences to the current conditions. The effects of
anchoring are pervasive and robust and are extremely difficult to ignore, even
when people are aware of the effect and aware that the anchor is ridiculous. Even
patently ridiculous anchor values have been shown to influence post-anchor estimates.
The effect of anchoring and adjustment grows with the size of the difference between
the anchor value – the original estimate provided – and the pre-anchor estimate – the
mean estimate people make before being exposed to an explicit anchor. In other words,
the bigger this difference, the greater the influence of the anchor value on the post-
anchor estimate.
Thus, for example, we could ask an estate agent to value a house without giving them
any clues (such as a listed price) or anchor values. The resulting estimate would be a
pre-anchor value. We could then tell them the listed price/anchor value and get them to
re-estimate the value of the house with this additional information. The resulting
estimate would be the post-anchor estimate.

The following example is taken from the paper
 Northcraft G B, and M A Neale , Experts, amateurs and real estate: an anchoring

and adjustment perspective on property pricing decisions, Organizational
behaviour and human decision processes, 39.
Example
An experiment was conducted in which a large number of real estate agents were asked
to value a property and come up with a recommended selling price. They were each
provided with an information booklet containing a large volume of information
concerning the property. The booklet was identical for all of the agents, except that
four different versions were used, each with a different listed (ie suggested) price for the
property. It turned out that the average selling price recommended by the agents
increased with the listed price as shown in the following table.
Listed price ($) Average recommended selling

price ($)
119,900 (Version 1) 117,745
129,900 (Version 2) 127,836
139,900 (Version 3) 128,530
149,900 (Version 4) 130,981
In the above example, the anchor value is the listed price. Thus, the agents’ estimates
were influenced by the “anchor” or benchmark given to them in the form of the listed
price. In addition it turns out that the further the anchor value gets from the “true”
value, then the more it will pull people’s estimates away from the true value.
3.3 Prospect theory
Prospect theory is a theory of how people make decisions when faced with risk
and uncertainty. It replaces the conventional risk-averse / risk-seeking decreasing
marginal utility theory based on total wealth with a concept of value defined in
terms of gains and losses relative to a reference point. This generates utility
curves with a point of inflexion at the chosen reference point.

Prospect theory attempts to explain why people may make asymmetric choices when
faced with similar possible gains and losses. It was originally developed by Kahneman
and Tversky in their paper:
 Kahneman, D and Tversky, A (1979), Prospect theory: an analysis of decision
under risk, Econometrica 47.
In this paper, Kahneman and Tversky describe an experiment in which they asked
people to choose between two alternatives:
 Alternative 1: an 80% chance of winning $4,000 and a 20% chance of winning

nothing
 Alternative 2: a 100% chance of winning $3,000.
Although the first alternative offers higher expected winnings ($3,200 v $3,000 for
certain), 80% of people chose Alternative 2. This choice is consistent with the
assumption of risk aversion that underpins expected utility theory. A risk-averse person
may prefer a more certain outcome, even if the expected gains are lower (because the
additional value derived from the extra certainty outweighs the additional value of the
higher possible return.
The same people were then offered the following choice:

 Alternative 3: an 80% chance of losing $4,000 and a 20% chance of losing
nothing
 Alternative 4: a 100% chance of losing $3,000.
Here 92% of people chose Alternative 3, even though the expected losses are greater
(expected losses of $3,200 v a certain loss of $3,000). This evidence suggests that
rather than being risk-averse, people may actually become risk-seeking when facing
losses.
Prospect theory suggests that:
 Value is based on gains and losses relative to some reference point.
 The loss in value from a loss is typically twice as much as the gain in value from
the same size monetary gain.
 People are typically risk-averse when considering gains relative to the reference
point and risk-seeking when considering losses relative to the reference point.

value placed on
gain or loss
losses gains
reference point
Figure 3.2: Prospect theory value function
Prospect theory suggests that the decision made depends on how a problem is presented
or “framed”, ie whether the available choices are presented as gains or losses relative to
the chosen reference point. If the alternative choices are presented as possible gains,
then the value function is concave, reflecting the risk-averse nature of individuals.
Conversely if they are presented as possible losses, then the value function is convex
reflecting the risk-seeking nature of individuals. So, the value function will be as
shown on the diagram above.
Prospect theory is therefore associated with the concept of framing.
3.4 Framing (and question wording)
The way a choice is presented (“framed”) and, particularly, the wording of a

question in terms of gains and losses, can have an enormous impact on the
answer given or the decision made. Changes in the way a question is framed of
only a word or two can have a profound effect.

Question 3.7
Which of the following two alternatives would you prefer?

1. $30 for certain plus a 50% chance of losing $20
2. $10 for certain plus a 50% chance of winning $20
In the same way, “structured response” questions are found to convey an implicit
range of acceptable answers.
In the book:
 Plous S, (1993), The psychology of judgement and decision making, McGraw-
Hill Inc
Plous describes an experiment in which people were asked the following questions
about the length of a film (the same one) they had all recently watched:
 Question 1: How long was the movie?
 Question 2: How short was the movie?
The mean answer to the first question was 2 hours and 10 minutes, whereas that to the
second question was 1 hour and 40 minutes!
3.5 Myopic loss aversion
This is similar to prospect theory, but considers repeated choices rather than a
single “gamble”. It may therefore be relevant when considering investment choices,
which can often be thought of as a series of repeated gambles. For example, if an
investor reviews and possibly changes its investment strategy on an annual basis, then
the investment strategy decision can be thought of as a series of repeated one-year
gambles.
Research suggests that investors are less “risk-averse” when faced with a multi-
period series of “gambles”, and that the frequency of choice / length of reporting
period will also be influential.
As its name suggests, myopic loss aversion relates to investors’ aversion to short-term
losses. The basic idea is that investors have been shown to be less “risk-averse” when
faced with a repeated series of “gambles” than when faced with a single gamble.

Thus, if the investor recognises that the investment strategy decision is in fact a series
of repeated short-term gambles and consequently takes a long-term view when
determining strategy, then they are likely to be less risk-averse than if they instead
consider only the immediate short-term gamble and so take too short-term a view. In
this latter case, they will tend to focus more on the short-term risk of loss than is
necessarily in their best interests, the consequence being that their resulting portfolio
ends up being overweight in less risky assets.
In addition, the extent of investors’ short-sightedness may be influenced by the

frequency with which they review their investment choices and/or the length of the
reporting period. For example, a pension fund that has to report its financial position
every year may be more averse to very short-term investment losses than one that has to
report only every three years.
Question 3.8
What might be the consequence on its investment strategy of requiring a pension fund
to report its financial position annually rather than triennially?
3.6 Estimating probabilities
In many situations, investors need to estimate probabilities, for example, in order to

assess the riskiness of a particular security. However, research again suggests that
investors’ estimates may be influenced by a number of biases.
Issues (other than anchoring) which might affect probability estimates include:
 Dislike of “negative” events – the “valence” of an outcome (the degree to

which it is considered as negative or positive) has an enormous influence
on the probability estimates of its likely occurrence.
In particular, experiments suggest that individuals are prone to underestimate the
probability that a negative event may occur.
 Representative heuristics – people find more probable that which they find
easier to imagine. As the amount of detail increases, its apparent likelihood
may increase (although the true probability can only decrease steadily).
As the amount of detail increases, the more specific (less generalised) the event
becomes and the less probable the occurrence of such a specific event must
become.

 Availability – people are influenced by the ease with which something can
be brought to mind. This can lead to biased judgements when examples of
one event are inherently more difficult to imagine than examples of another.
An example here concerns the incidences of deaths due to car crashes and
cancer. When asked to estimate the relative numbers of deaths due to each,
people tend to overestimate the number of deaths due to car crashes perhaps
because they receive more publicity and are easier to imagine.
3.7 Overconfidence
People tend to overestimate their own abilities, knowledge and skills. For example,
if you ask 100 people if they are better than average drivers, then you might not be
surprised if more than 50% of them reply “yes”!
Question 3.9
Do you think you will pass the Subject CT8 exam?
Moreover, studies show that the discrepancy between accuracy and

overconfidence increases (in all but the simplest tasks) as the respondent is more
knowledgeable! (Accuracy increases to a modest degree but confidence increases
to a much larger degree.) If this is true then it may not be wise to pass the Subject
CT8 exam!
Overconfidence could therefore be a potentially serious problem in fields such as

investment where most of the participants are likely to be highly knowledgeable.
Moreover, the available evidence suggests that even when people are aware that they
are overconfident they remain so.
This may be a result of:
 Hindsight bias – events that happen will be thought of as having been

predictable prior to the event; events that do not happen will be thought of
as having been unlikely prior to the event.
A possible example of the first type of event is the credit crunch of 2007/09. A
possible example of the second type of event is when an underdog is heavily
beaten in a sporting event. Although supporters may have had high hopes of an
upset prior to the event, after the event a heavy defeat will always have seemed
inevitable.

 Confirmation bias – people will tend to look for evidence that confirms their
point of view (and will tend to dismiss evidence that does not justify it).
For example, failing an exam might reinforce an existing preconception that the
exam system is a lottery – even if the real reason for failure was a lack of
preparation. In contrast, passing an exam might reinforce a view of the exam
system as a lottery, even if it was in fact due reward for months of long, hard
study!
3.8 Mental accounting
People show a tendency to separate related events and decisions and find it
difficult to aggregate events. Thus, rather than netting out all gains and losses (as
standard financial economic theory would suggest) people set up a series of “mental
accounts” and view individual decisions as relating to one or another of these
accounts.
Question 3.10
Consider the following scenarios:

 Scenario 1: You lose $10 today and find a $10 note on the street tomorrow.
 Scenario 2: Neither of the above events occurs.
Which scenario would you prefer?
An example of when mental accounting might be of relevance relates to the paying off
of a mortgage early. Consider the following two situations:
1. have a 100k mortgage and 10k of savings in the bank
2. use 5k of savings to pay off part of your mortgage, so that you end up with a 95k
mortgage and 5k of savings in the bank.
In theory (ie ignoring issues such as tax and the interest rates applicable) you should be
indifferent between these two situations because either way you have a net indebtedness
of 90k (and somewhere to live!).
In practice, however, this may not be the case. Some people may actually feel happier
in the second situation because they feel that a smaller mortgage means that they have
less debt. This will be the case if they use separate mental accounts for savings and
debt.

In practice, of course, the choice of course of action will be driven to some extent by
issues such as tax, the interest rates applicable and the need to hold some liquid assets.
3.9 Effect of options
Experimental evidence also suggests that the range of options or choices presented to
people may influence their choices.
In addition to the “framing” effect discussed in Section 3.4 above, other issues
include:
 the primary effect – people are more likely to choose the first option
presented
 the recency effect – in some instances, the final option that is discussed
may be preferred! The gap in time between the presentation of the options
and the decision may influence this dichotomy.
More specifically, the sooner/later the decision is made, the more likely it is that
the first/last option will be chosen.
 other research suggesting that people are more likely to choose an
intermediate option than one at either end!
 a greater range of options tends to discourage decision-making. On the
other hand, a higher probability is attributed to options explicitly stated than
when included in a broader category.
 status quo bias – people have a marked preference for keeping things as
they are.
 regret aversion – by retaining the existing arrangements, people minimise
the possibility of regret (the pain associated with feeling responsible for a
loss).
 ambiguity aversion – people are prepared to pay a premium for rules.
Much standard financial economic theory (such as portfolio theory, which is the
subject of Chapter 5) assumes that investors know the actual distribution of
future investment returns when making their investment choices – although they
don’t know what the actual investment return will turn out to be in any future
period. The terms ambiguity and uncertainty are both used to refer to a situation
in which the investor is uncertain about the distribution of future investment
returns. An investor who dislikes such ambiguity will tend to err on the side of
caution when making an investment choice. The investor will also be prepared
to pay for further information that reduces the degree of uncertainty faced.

Question 3.11
List the eight common themes found in research on behavioural finance.
3.10 Further reading
The material in this section is based on the paper:

 Nigel Taylor (2000), Making actuaries less human: lessons from behavioural
finance, Staple Inn Actuarial Society (SIAS).
This is an easy-to-read paper, which is well worth studying if you are interested in these
ideas. It can be downloaded from the SIAS website: www.sias.org.uk.

4 Exam-style question
We finish this chapter with an exam-style question on stochastic dominance.
Question
(i) Define first-order and second-order stochastic dominance. Illustrate the

definitions by sketching cumulative distribution functions of two random
variables which represent the returns on two investments, one of which
dominates the other. [5]
(ii) Explain how an investor’s economic characteristics will affect his choice of an
investment that:
(a) first-order stochastically dominates another
(b) second-order stochastically dominates another. [3]
[Total 8]

Solution
This is Subject 109, September 2000, Question 9.
(i) Definitions
A distribution of investment returns A is said to exhibit first-order stochastic

dominance over a distribution of investment returns B if:
FA ( x) £ FB ( x) for all x, and

FA ( x) < FB ( x) for some value of x.
In other words, the probability of B producing a return below a certain value is never
less than the probability of A producing a return below the same value and exceeds it
for at least some value of x.
F(x)
1
FB ( x ) FA ( x )
FB ( y )
FA ( y )
0 y x
Here FA ( x) must never be above FB ( x) . Note, however, that the lowest possible value
of x may be non-zero and even negative.
A distribution of investment returns A is said to exhibit second-order stochastic

dominance over a distribution of investment returns B if:
x x
 a
FA ( y ) dy  
a
FB ( y ) dy
for all x, with the strict inequality holding for some value of x, where a is the lowest
return that the portfolios can possibly provide.

F ( x)
FA ( x)
FB ( x)
0 a x x
Here the area under FB ( x) must never be less than the area under FA ( x) for any value
of x.
(ii) Economic characteristics
(a) An investor who prefers more to less – ie who is not satiated – will always
choose an investment whose distribution of returns first-order stochastically
dominates another.
(b) An investor who prefers more to less and is also risk-averse will always choose
an investment whose distribution of returns second-order stochastically
dominates another.

Chapter 3 Summary
Stochastic dominance
Stochastic dominance offers an approach to modelling choices under uncertainty that

does not require the use of explicit utility functions.
Given two investment portfolios, A and B, with cumulative probability distribution

functions of returns FA and FB respectively:
The first-order stochastic dominance theorem states that A will be preferred to B if:
 the investor prefers more to less, U ( x)  0 and
 FA ( x)  FB ( x) for all x, with FA ( x)  FB ( x) for at least one x.
The second-order stochastic dominance theorem states that A will be preferred to B if:
 the investor prefers more to less, U ( x)  0 ,
 he is risk-averse, U ( x)  0 , and

x x
 a
FA ( y ) dy   FB ( y ) dy, for all x, with the strict inequality holding for at least
a
one x.
Behavioural finance
The field of behavioural finance relates to the psychology that underlies and drives
financial decision-making behaviour. Eight common themes include:
1. anchoring and adjustment
2. prospect theory
3. framing (and question wording)
4. myopic loss aversion
5. estimating probabilities
6. overconfidence
7. mental accounting
8. effect of options.

This page has been left blank so that you can keep the chapter summaries
together for revision purposes.

Solution 3.5
cumulative
probability
1
Asset U
A
Asset V
B
6 7 8 9 return, %
Figure 3.4 – Assessing the stochastic dominance of Assets U and V
In this case, first-order stochastic dominance is insufficient to choose between the

assets, because the cumulative probability graphs cross. U is second-order
stochastically dominated by V because the extra possibility of obtaining 8%
(represented by the box marked A) is of less value to the investor than the possibility of
avoiding 6% (represented by Box B).
Solution 3.6
The main advantage of stochastic dominance is that it does not require explicit
formulation of the investor’s utility function, but can instead be used to make
investment decisions for a wide range of utility functions.
The main disadvantages are that it:

 may be unable to choose between two investments and
 generally involves pair-wise comparisons of alternative investments, which may
be problematic if there is a large number of investments between which to
choose.

Solution 3.7
The two alternatives are of course identical in that they both offer an equal chance of
winning either $30 or $10. However, experimental evidence based on similar choices
suggests that people can and do view identical alternatives differently depending on
how they are framed or worded. Thus, the proportion of people selecting each choice
typically differs greatly from 50%.
Solution 3.8
The consequence of requiring a pension fund to report its financial position annually
rather than triennially might be to force the fund to invest in less volatile assets in order
to reduce the risk of having to report a poor financial position. It might therefore force
the fund to invest less heavily in equities, and possibly also less heavily in long-term
bonds.
This assumes that assets are valued using market values.
Solution 3.9
Only you will know how you answered this. However your answer probably reflected
behavioural issues as much as an objective assessment of your true chances. This is not
an entirely satisfactory example as some people may in fact be under-confident as
regards their chances of passing the exam. Either way, it is quite possible that the
percentage of you answering “yes” is very different to the 50% or so which may turn
out to be the case in practice!
Solution 3.10
Standard financial economic theory suggests that you should be indifferent between the
two outcomes because they both result in no net gain or loss. In practice, however, you
might have chosen one or the other. When asked about scenarios of this nature in
experiments, there is typically a very strong majority preference for one or other of the
scenarios described.

Solution 3.11
The eight common themes found in research on behavioural finance are:

1. anchoring and adjustment
2. prospect theory
3. framing (and question wording)
4. myopic loss aversion
5. estimating probabilities
6. overconfidence
7. mental accounting
8. effect of options.





the course.

CT8-10: Stochastic models of security prices Page 21
4 The Wilkie model
4.1 Introduction
We will now look briefly at the Wilkie model, which is an example of an autoregressive
model that can be used to model various economic variables.
Although it is rarely used in its published form, the Wilkie model (full details of
the Wilkie model are set out in his paper “More on a Stochastic Asset Model for
Actuarial Use”, BAJ Volume 1, pages 777–964) provides a convenient published
example of a model designed explicitly for long-term applications.
David Wilkie has also written more recent papers relating to his model.
4.2 Structure of the Wilkie model
The Wilkie model is used to simulate the future values of economic variables in each
future year, over a number of years.
The structure of the Wilkie model as originally conceived can be shown

diagrammatically as a “cascade” model. This form shows how the parameters of
the model can be estimated serially, but also “hides” some of the implicit
structure (in particular the lack of correlation between some variables) imposed
on the relationships between the variables.
Retail price inflation
Share dividends Share dividend yield Wage inflation
etc.
(This diagram is part of the Core Reading.)
You can see from the diagram that, for example, the rate of inflation in a particular year
will feed into the values of the dividend yield for the next year but not vice versa. More
generally, under this cascade structure, higher variables influence those lower down but
not vice versa, ie the causality “cascades” down through the model.

Page 22 CT8-10: Stochastic models of security prices
The “etc” in the diagram indicates that this is just a subset of the full model, which
includes many other variables, and that there are further tiers of variables below the
ones shown, eg dividend growth, share prices, interest rates (short- and long-
term), property returns, index-linked stock and currency exchange rates.
The Wilkie model can also be recast as a single multivariate model based on a
vector autoregressive moving average (VARMA) process. The vector
representation makes it clear that the parameter count is far higher than is first
apparent as a zero value is a parameter value.
In the vector representation, rather than modelling each variable individually, with its
own equation, we instead put the variables of interest into a vector and model the vector
as an ARMA process, ie:
X t = A + BX t -1 + C0 Zt + C1 Zt -1 + C2 Zt - 2 .
where:
 X t is a vector of this period’s values of the variables of interest, eg inflation, the

share dividend yield.
 X t -1 is a vector of last period’s values of the same set of variables.
 Zt -i , i = 0, 1, 2 , are vectors of random innovation terms (each with a standard

normal distribution).
 A, B, C0 , C1 , C2 are all matrices of constants, estimated using historical data.
In fact, the Wilkie model turns out to be a VARMA(1, 2) process, which means it is:
 autoregressive of order 1 or AR(1), as this year’s vector of variable values is

assumed to depend upon last year’s vector of variable values (but not values
before then)
 moving average of order 2 or MA(2), as this year’s vector of variable values is

assumed to depend upon the random innovation terms from this year, last year
and also from two years ago.

In theory, it would be possible to set up the model so that each of the variable values
this year is directly dependent on each of the variable values last year, which would
result in a set of equations with a large number of parameters to which we could assign
values. In practice, however, Wilkie assumed a much simpler cascade structure, which
drastically reduces the number of parameters actually used. For example, in the Wilkie
model, inflation this year is linked only to inflation last year, which is equivalent to
saying that the parameters for the links to last year’s values of all the other variables are
equal to zero.
In fact, Wilkie models price inflation as an AR(1) process using the equation:
I t  QMU  QA  I t 1  QMU   QSD. QZt
Here:
 I t , the force of inflation, is derived from movements in the Retail Price Index
Qt , so that I t = ln Qt - ln Qt -1
● QMU , QA and QSD are parameters to be estimated
● QZt is a series of independent, identically distributed (iid) standard normal

random variables, also called “innovations”.
Notice the structure of the equation:
This year’s value = long-run mean + QA ¥ (last year’s value - long-run mean)
+ a “random shock to the system”
In this structure:
● the long-run mean is QMU
● the auto-regressive component is based upon the difference between last year’s
value and the long-run average value of inflation, ie QA [ I t -1 - QMU ] . As
QA < 1 , the process is also mean-reverting.
● the “shock to the system” (random component) is QSD . QZt . QSD dictates the
typical size of this “shock”.

Also, none of the other variables in the model (eg dividend streams and dividend yields)
appear in the equation because inflation is assumed to influence them, but they are
assumed not to influence inflation.
Question 10.11
What do you think might be an appropriate value for QMU ?
The original form was decided upon using a combination of high-level economic
expectations, statistical analysis and the requirement to keep the model
parsimonious, ie to avoid the estimation of “too many” parameters. It should be
remembered that there is no over-riding economic reason for the model to have
the particular form it was originally published in. Analysis using different
(for example more recent) data sets may indicate that the best statistical model
has a different form. This may also be true when analysing economies other
than the UK.

7 Exam-style question
We finish this chapter with an exam-style question based on the lognormal model of
share prices.
Question
An investor has decided to model PPB plc shares using the lognormal model. Using
historical data, she has estimated the drift and volatility parameters to be 6% and 25%
respectively. PPB’s current share price is $2.
(i) Calculate the mean and variance of PPB’s share price in one year’s time. [3]
(ii) Calculate the probability that:
(a) PPB’s shares fall in value over the next year.
(b) PPB’s shares yield a return of greater than 30% over the next year.
Assume that no dividends are to be paid over the next year. [4]
[Total 7]

Solution
(i) Mean and variance of PPB’s share price in one year’s time
Based on the information given in the question:
log S1  log S0 ~ N   ,  2 
 
So, using the formulae for the mean and variance of the lognormal distribution from
page 14 in the Tables, we have:
(
E ÈÎ S1 S0 ˘˚ = S0 exp m + ½s 2 )
( )( ( ) )
Var ÈÎ S1 S0 ˘˚ = S02 exp 2 m + s 2 exp s 2 - 1
So, with m = 0.06, s = 0.25 and S0 = 2 we have:
( )
E ÈÎ S1 S0 ˘˚ = 2 ¥ exp 0.06 + ½ ¥ 0.252 = 2.1911
( )( ( ) )
Var ÈÎ S1 S0 ˘˚ = 22 ¥ exp 2 ¥ 0.06 + 0.252 exp 0.252 - 1 = 0.30963
(ii)(a) Probability that PPB’s shares fall in value over the next year
Here we want the probability that S1 < S0 .
ÈS ˘
P [ S1 < S0 ] = P Í 1 < 1˙
Î S0 ˚
È ÊS ˆ ˘
= P Í log Á 1 ˜ < log (1)˙
ÎÍ Ë S0 ¯ ˚˙
È Ê S1 ˆ ˘
Í log Á ˜ - 0.06 ˙
Í Ë S0 ¯ 0 - 0.06 ˙
=P <
Í 0.25 0.25 ˙
Í ˙
ÎÍ ˙˚

ie P [ S1 < S0 ] = P [ Z ~ N [0, 1] < -0.24]
= 1 - F(0.24)
= 1 - 0.59483
= 0.40517
So, there is a probability of almost 41% that the share price will fall over the next year.
(ii)(b) Probability that PPB’s shares yield more than 30% over the next year
Here we want the probability that S1 > 1.30 S0 .
ÈS ˘
P [ S1 > 1.30 S0 ] = P Í 1 > 1.30 ˙
Î S0 ˚
È ÊS ˆ ˘
= P Í log Á 1 ˜ > log (1.30)˙
ÍÎ Ë S0 ¯ ˙˚
È Ê S1 ˆ ˘
Í log Á ˜ - 0.06 ˙
Í Ë S0 ¯ log (1.30) - 0.06 ˙
=P >
Í 0.25 0.25 ˙
ie Í ˙
ÍÎ ˙˚
= P [ Z ~ N [0, 1] > 0.80946]
= 1 - F(0.80946)
= 1 - 0.79087
= 0.20913
So, there is a 21% or so chance that the share will yield a return of 30% or more over
the next year.
The relatively high probabilities in both (a) and (b) reflect the volatility parameter of
25%, which isn’t unrealistic for an individual share.





the course.

CT8 Pu 14

Uploaded by

Copyright:

Available Formats

CT8 Pu 14

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CT8 Pu 14

Uploaded by

Copyright:

Available Formats

CT8: CMP Upgrade 2013/14 Page 1

This CMP Upgrade contains:

 all changes to the Syllabus objectives and Core Reading

The Actuarial Education Company © IFE: 2014 Examinations

1 Changes to the Syllabus Objectives and Core Reading

1.1 Syllabus Objectives

A new objective has been added as follows:

(i) 9. Discuss the key findings in behavioural finance.

1.2 Core Reading

© IFE: 2014 Examinations The Actuarial Education Company

2 Changes to the ActEd Course Notes

The Actuarial Education Company © IFE: 2014 Examinations

The following equation in the middle of the page in terms of St :

has been replaced by the following equation in terms of St :

© IFE: 2014 Examinations The Actuarial Education Company

3 Changes to the Q&A Bank

(i) Explain what is meant by the Wilkie model

It can therefore be used to simulate the possible future development of investment

It has a cascade structure. [½]

The key variable is the force of inflation. [½]

This is modelled as a first-order autoregressive process with normally distributed

It is therefore a particular case of a vector autoregressive moving average (VARMA)

The Actuarial Education Company © IFE: 2014 Examinations

(iii) Two key differences between the models

(i) Explain fully overconfidence

© IFE: 2014 Examinations The Actuarial Education Company

This may be a result of:

 Hindsight bias – events that happen will be thought of as having been

The Actuarial Education Company © IFE: 2014 Examinations

(iii) A financial advisor is due to present a range of possible investment strategies to

(i) Define framing

Framing is related to:

 anchoring and adjustment – as the answer to a question will tend to be anchored

 prospect theory – as decisions will be influenced by how information is

© IFE: 2014 Examinations The Actuarial Education Company

(ii) How to present the past performance of the fund favourably

She will need to decide:

The Actuarial Education Company © IFE: 2014 Examinations

A zero-coupon bond (ZCB) promises to pay a return of 8% when it matures in a year’s

A 1-year risk-free government ZCB provides a return of 5%.

A bond analyst is going to use a two-state intensity-based model to derive risk-neutral

(i)(a) Downside semi-variance

The expected return on the risky ZCB is equal to:

0.9  8%  0.05  6%  0.05  0%  7.5% [1]

Consequently, the downside semi-variance is given by:

 7.5  6 2  0.05   7.5  0 2  0.05  2.925%% [1]

(i)(b) Shortfall probability relative to risk-free rate

© IFE: 2014 Examinations The Actuarial Education Company

(i)(c) Expected shortfall conditional on a shortfall occurring

(ii)(a) General risk-neutral pricing formula for a ZCB

(ii)(b) Derive formula for risk-neutral transition intensity function

 exp     ( s )ds      exp     ( s )ds 

The Actuarial Education Company © IFE: 2014 Examinations

Differentiating both sides with respect to T then gives:

(i) Short rate model of interest rates

© IFE: 2014 Examinations The Actuarial Education Company

r(t) itself is usually assumed to be a diffusion or Ito process, with a stochastic

(ii) Explain what is meant by a one-factor short rate model

The single source of randomness is typically assumed to be a standard Brownian motion