lOMoARcPSD|4964459

Exam 2010, answers

Derivatives and Risk Management (The University of Warwick)

Scan to open on Studocu

Studocu is not sponsored or endorsed by any college or university

Downloaded by bijay agrawal ([email protected])
 lOMoARcPSD|4964459

THE UNIVERSITY OF WARWICK

SUMMER EXAMINATION 2010

IB 3590 DERIVATIVES AND FINANCIAL RISK MANAGEMENT

SOLUTION NOTES

Remarks:

The structure of this exam was changed slightly relative to last year’s paper, partly in
response to comments from the external examiner, but mainly because of the dramatic
increase in student numbers (from just over 100 to almost 300): in view of the very
tight marking deadlines, we felt it was necessary to shift the emphasis slightly towards
the quantitative questions (Section A), and have only one essay-style question.

Specifically, students must answer 3 of the 4 (quantitative) questions in Section A,

and any one of their choice of the 4 questions from Section B. All questions from
both sections carry equal weight (each counting 25% towards the exam mark).

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

SECTION A

Question 1 – Solution Notes:

Part a):

With a yield of 5% and annual coupons of 8%, the value of bond A is:

VA = exp(–0.05)×8% + exp(–0.05×2)×8% + exp(–0.05×3)×108% = 107.80%.

As B is a zero-coupon bond, its value is VB = exp(–0.05×5)×100% = 77.88%.

As Bond B is zero-coupon, duration must be equal to maturity, i.e. DB = 5. For Bond

A, we use the standard duration formula with continuously compounded yields (given
on the formula sheet):

DA = exp(–0.05)×(8% / VA)×1 + exp(–0.05×2)×(8% / VA)×2

+ exp(–0.05×3)×(108% / VA)×3 = 2.792.

Parts b) and c):

For Scenario (1), the answers for Parts b), c), and e) are identical: if yields do not
change, we know that we will earn the yield (this was shown in class). As Claire has
invested $100 million, her future value in 4 years will be:

100 million × exp(0.05×4) = 122.14 million.

Scenario (2), Bond A (Part b):

We first work out the future value (at the end of year 4) of the bond’s cash flows,
assuming these can be re-invested at the new rate of 4%:

VA(4; 4%) = exp(0.04×(4 – 1))×8% + exp(0.04×(4 – 2))×8%

+ exp(0.04×(4 – 3))×108% = 130.09%.

As the value of the bond was VA = 107.80% at the time of purchase (part a), Claire
would have bought 92.76 million (= 100 million / 1.0780) par of Bond A. The total
value 4 years from now is hence 92.76 million × 1.3009 = 120.68 million. This is
about 1.2% (almost $1.5 million) less than in Scenario (1).

Scenario (2), Bond B (Question c):

Again, we first compute the value (at the end of year 4) of bond B. As this is a zero-
coupon bond, we just need to discount the redemption value from year 5 to year 4:

VB(4; 4%) = exp(–0.04)×100% = 96.08%.

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 1 – Solution Notes (Cont’d):

As the value of the bond was VB = 77.88% at the time of purchase (part a), Claire
would have bought 128.40 million (= 100 million / 0.7788) par of Bond B. The total
value 4 years from now is hence 128.40 million × 0.9608 = 123.37 million. This is
about 1% (about $1.2 million) more than in Scenario (1).

Part d):

We need to find the “weights” (i.e. % of money invested) for the 2 bonds so that the
aggregate duration is = 4. As we know DA = 2.792 and DB = 5, we have to solve:

wA × 2.792 + (1 – wA) × 5 = 4

This implies wA = 45.28%, and hence Claire should invest $45.28 million in Bond A,
and the remaining $54.72 million in Bond B.

Given the current prices of the bonds (from Part a), Claire should buy $42.01 million
par of Bond A, and $70.26 million par of Bond B.

Part d):

Scenario (1): if rates are unchanged, Claire will have the same as in Parts a) and b),
$122.14 million (this should be obvious, but can also be checked easily).

Scenario (2): if rates have fallen to 4%, we need to compute the future values of the
cash flows we receive from Bond A, and the present value of the par redemption from
Bond B. We know from parts b) and c) that the future values of the two bonds are
130.09% and 96.08%, respectively. As Claire holds $42.01 million par of Bond A,
and $70.26 million par of Bond B, the year-4 value of the portfolio is:

42.01 million × 130.09% + 70.26 million × 96.08% = 122.15 million.

This is only about $10,000 (less than 1 basis point!) off the value in Scenario (1).
Note that because of convexity, Claire would have made a similar gain even if rates
had moved the other way (e.g. increased to 6%). The results show that, for relatively
small lateral movements in rates, immunization works very well. Of course, if the
term structure of interest rates changes shape, it will no longer work. For example, if
the term structure steepens, Claire will incur a loss even on the immunized portfolio.

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 2 – Solution Notes:

Part a):

We denote by S and F the current price per unit of the underlying exposure and the
futures, respectively, and by ST and FT the prices at the futures contract’s maturity. If
the “size” of our exposure to S is “Q”, and the size of the (short) hedge is “–H”, the
value of our position at the target date is:

π = Q × [ S + (ΔS – h×ΔF) ], where h = H/Q is the “hedge ratio”.

We compute the variance of this as:

Var(π) = Q2 × [ σS2 + h2 × σF2 – 2h × σS,F ]

The FOC to obtain the hedge ratio that minimizes the variance is hence:

0 = Q2 × [ 2h × σF2 – 2 × σS,F ], which implies h = σS,F / σF2

This can of course be re-formulated in terms of correlation: h = ρS,F × (σS / σF).

Part b):

Here, if we define S as the value of the entire portfolio (i.e. $12 million in March),
then the “size” of the exposure is Q = 1. With the given data, the hedge ratio is:

h = 0.8 × ($600,000 / 60) = $8,000

As this is the number of “units” of the index, but each contract has a “point value” of
$250, we need to short $8,000 / $250 = 32 contracts. Note that the nominal value of
the hedge is 32 × $250 × 1,200 = $9.6 million, or 80% of the original exposure.

Part c):

Using the standard “cash-and-carry” pricing equation for forward prices (discussed at
great length in the lectures and seminars), we get the prices of the June contract:

March April May June

S&P 500 1,200 1,180 1,280 1,160
June Futures 1,209.00 1,185.90 1,283.20 1,160.00

As an example, we show the workings for the valuation in April:

F = 1,180 × [ 1 + (5% – 2%) × (2 × 30 / 360) ] = 1,185.90.

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 2 – Solution Notes (Cont’d):

Part d):

Initially (in March) we need to deposit $30,000 margin per contract, and as we are 32
contracts short, that’s $960,000.

In April, the futures price has dropped from 1,209 to 1,185.9, and since we’re short
we gain –32 × $250 × (1,185.9 – 1,209) = $184,800.

By May, the futures price has increased to 1,283.20, which brings the balance of our
margin account down to $366,400. This is below the maintenance margin (which,
given that we have a position of 32 contracts, is 32 × $20,000 = $640,000). So we get
a margin call, and have to deposit an additional $593,600 (= $960,000 – $366,400) to
get back to the initial margin.

Finally, in June the futures price is down again at 1,160, so our short position makes a
gain, and the balance of the margin account increases to $1,945,600.

Change Transaction Balance Notes

March + 960,000 960,000 Initial deposit
April + 184,800 1,144,800 Marking-to-market
May – 778,400 366,400 Marking-to-market
+ 593,600 960,000 Margin call!
June + 985,600 1,945,600 Marking-to-market

Part e):

In March, the portfolio had a value equal to exactly 10,000 times the index level. If
the portfolio has followed the index, the same should be true in June: the value will be
down by $400,000 to $11.6 million (a loss of (1/30)%). Peter’s total P&L on the
hedged position would hence be:

Loss on portfolio – 400,000

Initial margin deposit – 960,000
Margin call – 593,600
Return of margin balance + 1,945,600

This adds up to a net loss of $8,000. Why do we make a loss? Due to the imperfect
correlation between the index and the portfolio, the “optimal” hedge covers only 80%
of the initial exposure. But in the scenario considered in part e), the portfolio turned
out to move in perfect unison with the index, so we’re slightly under-hedged and left
with a net loss.

But, if portfolio and index move perfectly together, and we have hedged only 80% of
our exposure (see part b), shouldn’t the loss be 20% of the fall in portfolio value (i.e.
$80,000)? In a nutshell, our hedge was indeed 80% relative to the initial index value,
but we also earn the “carry” on our futures position (and we have ignored the cost of
funding the margin!).

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 3:

Part a):

The complete binomial model is shown below:

(2,2) S C P
value 37 12 0

(1,1) S C P
value 31 6.125 0.000 payoff from (1,1) 12.000 0.000
delta 1.000 0.000
(0,0) S C P cash 25.000 0.000 (2,1) S C P
value 25 2.992 2.743 payoff from (0,0) 6.125 0.000 value 27 2 0
delta 0.531 ‐0.469
cash 10.329 ‐14.546 (1,0) S C P payoff from (1,1) 2.000 0.000
net cash flow 0.00843 ‐0.0428 value 21 0.817 4.692 payoff from (1,0) 2.000 0.000
delta 0.200 ‐0.800
cash 3.400 ‐21.600 (2,0) S C P
payoff from (0,0) 0.817 4.692 value 17 0 8

payoff from (1,0) 0.000 8.000

Figure 3.1: Binomial Model for Question 3

As up- and down moves are equally likely, it is obvious that nodes (1,0) and (1,1)
each occur with probability 50%. For the final stage, there are 2 “paths” leading to
the “middle” node (2,1), but only one path each leading to (2,0) or (2,2), and hence
the probabilities of the final stock prices are 25%, 50%, and 25%, respectively.

Part b):

The results are shown in Figure 3 above. We give only one set of calculations here,
all others are similar. In nodes (2,2) and (2,1), the call option is in-the-money, and
pays off $12 (= $37 – $25) or $2 (= $27 – $25), respectively.

We can thus calculate the “delta” in node (1,1):

δ(1,1) = ($12 – $2) / ($37 – $27) = 1.

This should be no surprise as the call is in-the-money in both “successor” nodes (2,2)
and (2,1). The matching cash position is β(1,1) = δ(1,1)×$37 – $12 = $25. Using the
one-period binomial option pricing equation, we get the value of the call in (1,1) as:

C(1,1) = δ(1,1) × $31 – PV(β(1,1)) = $6.125.

Here, we used the fact that the one-period discount factor is exp(–0.25×0.02) = 0.995.
Continuing this process by “backwards induction”, we get the “fair” value of the call
option as of today (the “root” of the model) as C = $2.992.

Using the same methodology, the value of the put option is P = $2.743.

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 3 – Solution Notes (Cont’d):

We check the put-call parity:

P + S = $2.743 + $25 = $27.743,

C + PV(Strike) = $2.992 + (0.995)2×$25 = $27.743.

No surprise here, as the binomial pricing formula is derived from the principle of
absence of arbitrage …

Part c):

First we check for violations of put-call parity:

Assuming hypothetically that the call is over-priced relative to the put, then we’d sell
the call, buy the put and the stock, and borrow PV($25) at 2%. By design, pay-offs at
maturity will cancel out. As we are writing the call, we receive the bid price ($3.00),
and we pay at the ask ($2.80) for the put. The total initial cash flows would be:

+ $3.00 – $2.80 – $25.00 + PV($25) = – $0.05 (no arbitrage!)

Similarly, if we buy the call and write the put, initial cash flows would be:

– $3.10 + $2.70 + $25.00 – PV($25) = – $0.15 (no arbitrage!)

In other words, even though the options are mis-priced, we cannot realize arbitrage
profits by a static put-call strategy as they are wiped out by the bid-ask spread.

We thus have to investigate delta-based strategies. The spread of the put brackets the
model price ($2.70 < $2.743 < $2.80), so no profits here. But the call is over-priced
even at the ask ($3.00 > $2.992). So, the strategy is: write the call at $3.00, and then
implement the replicating strategy using the stock and risk-free asset.

Part d):

(1)

We write the call and collect $3.00. The replicating strategy can be derived from
Figure 3.1: at each node, we hold “delta” units of the stock, and an amount of face
value “cash” in the risk-free asset. For example, we start off the strategy by buying
0.531 units of the stock, and borrow PV($10.329) for repayment next period. (Of
course, in reality we’d have to gear this up as we cannot trade fractional units, so we
could for example write 10 call contracts on 1,000 shares, and buy 531 shares).

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 3 – Solution Notes (Cont’d):

(2)

At the initial node, net cash flows (per share) are:

+ $3.00 – 0.531 × $25 + PV($10.329) = + $0.008

This is exactly the amount of mi-pricing relative to the ask price of the call (note, the
above is the precise amount coming from my Excel spreadsheet, students may get a
little rounding error here, but it should still be positive).

At node (2,2), we hold 1 unit of the stock (bought at node (1,1)) which is now worth
$37, and we have to pay back $25 on the loan taken out in the preceding node. The
net pay-off is hence $37 – $25 = $12, which exactly offsets the payoff to the call
option, so the total pay-off is zero.

At node (2,1), we’re either coming from (1,1), or from (1,0). Coming from (1,1), we
hold 1 unit of stock and have to repay $25, and since the stock is now worth $27, this
is a total of $2. Coming from (1,0), we hold 0.2 units of stock and have to pay back
$3.40, which gives 0.2 × £27 – $3.40 = £2. In both cases, the stock-and-cash strategy
matches the payoff to the option.

At node (2,0), we’re coming from (1,0) as above (holding 0.2 units of stock and
having to pay back $3.40), but now the stock is worth only $17, so the that the pay-off
of the strategy is 0.2 × $17 – $3.40 = 0. In all cases, the strategy in stock and cash
perfectly offsets the payoff to the short call.

(3)

In node (0,0) (the “root of the model”), we buy 0.531 units of stock, and borrow for
repayment of $10.329 next period.

At node (1,1), with a stock price of $31, we thus have 0.531×$31 - $10.329 = $6.13.
Going forward, we need to buy 1 unit of stock, and borrow PV($25). This will cost us
$31 – PV($25) = 6.13$. Exactly what we have coming in.

At node (1,0) the stock price is $10 lower than at (1,1), so we have $5.31 less. Going
forward, we now need to buy 0.2 units of stock and borrow PV($3.40). This will cost
0.2 × $21 – PV($3.40) = $0.82 (exactly $5.31 less than the $6.13 in (1,1)).

In both nodes (1,1) and (1,0), the pay-off of the strategy formed in node (0,0) exactly
matches the cost of the strategy going forward, i.e. the strategy is “self-financing”!

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 4 – Solution Notes:

Part a):

1-year rate:

For the 1-year discount factor, we just use the money market rate R1 = 3.050%:

B1 = 1 / (1 + R1) = 1 / 1.03050 = 0.9704

This implies a continuously compounded 1-year rate of r1 = – ln(0.9704) = 3.00%.

2-year rate:

We know that the 1-into-2-year forward rate is F1,2 = (B1 / B2 – 1). As we know the
forward rate is F1,2 = 4.288%, and we have B1 = 0.9704 from before, we can solve:

B2 = B1 / (1 + F1,2) = 0.9704 / 1.04288 = 0.9305

This implies a continuously compounded rate of r2 = – ln(0.9305) / 2 = 3.60%.

3-year rate:

The price of the 3-year 6% bond is 105.70. We know this must equal the discounted
present value of the bond’s future cash flows, and hence we can write:

105.70 = (B1 + B2) × 6 + B3 × 106

Since we know B1 = 0.9704 and B2 = 0.9305 from before, we can now solve the
above for B3 = [105.70 – (0.9704 + 0.9305) × 6] / 106 = 0.8896. This implies a
continuously compounded rate of r3 = – ln(0.8896) / 3 = 3.90%.

4-year rate:

We know that the 4-year swap rate can be written as X4 = (1 – B4) / (B1 + … + B4).
Re-arranging this to solve for B4, we get

B4 = [1 – X4 × (B1 + B2 + B3)] / (1 + X4)

With the discount factors B1, B2, and B3 calculated before, and the 4-year swap rate of
X4 = 4.158%, we get B4 = 0.8487. This implies a continuously compounded 4-year
rate of r4 = – ln(0.8487) / 4 = 4.10%.

5-year rate:

With the 5-year zero-yield of Y5 = 4.289%, we can get calculate the discount factor:

B5 = 1 / (1 + Y5)5 = 0.8106

This implies a continuously compounded 5-year rate of r5 = – ln(0.8106) / 5 = 4.20%.

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

Question 4 – Solution Notes (cont’d):

Part b):

With annual frequency, the “t-into-(t+1)” forward rate (simple compounding) is:

Ft,t+1 = Bt / Bt+1 – 1

Of course the “0-into-1” forward rate is equal to the 1-year spot rate, F0,1 = 3.050%.
For the other forward rates, we use the discount factors calculated in Part a):

F1,2 = 4.288% (= 0.9704 / 0.9305 – 1)

F2,3 = 4.598% …
F3,4 = 4.819%
F4,5 = 4.700%

Part c):

We know (as shown in the lectures) that to value the swap, we can simply replace the
floating payments by payments at the respective forward rates. As Tom is paying
fixed at 4% and receiving floating, the value VS of the swap to Tom is the sum of the
present values of the cash flows (Ft,t+1 – 4%) × $1 million (t = 0, 1, … 4):

VS = $1 million ×[B1 × (F0,1 – 4%) + B2 × (F1,2 – 4%) + … + B5 × (F4,5 – 4%)]

= – $11,400 (rounded to the nearest $100)

In other words, at 4% fixed rate, the swap is “below market” and, as Tom is paying
fixed, has positive value from his perspective. Tom will have to pay an up-front
premium of $11,400 to the counter-party to make the deal “fair”.

Part d):

As in Part a), we know from the lectures that the 5-year swap rate can be written as:

X5 = (1 – B5) / (B1 + … + B5).

Using the discount factors calculated in Part a), we get X5 = 4.256%.

Downloaded by bijay agrawal ([email protected])

 lOMoARcPSD|4964459

SECTION B

Questions 5 … 8:

I hesitate to provide model answers for these questions, because I want students to be
creative here. The bullet points in the questions themselves give some guidance, but
otherwise I think it should be a “free-style” exercise.

Downloaded by bijay agrawal ([email protected])