Inverse Floater Valuation Parameters
Inverse Floater Valuation Parameters
Inverse Floater Valuation Parameters
In this section, we will show the valuation of callable inverse floater issued by UBS
AG, callable cumulative inverse floater issued by Royal Bank of Scotland Plc, and
callable daily range accrual note issued by UBS AG respectively. The least-squares
Monte Carlo simulation of lognormal forward LIBOR is applied to all these three
cases.
Issuer:
UBS AG
100.00%
Issue Date:
04 October 2004
Maturity Date:
04 October 2007
44
Interest Amount:
Period
Coupon Rate
Oct04 Apr05
Apr05 Oct05
Oct05 Apr06
Apr06 Oct06
Oct06 Apr07
Apr07 Oct07
30/360
Early Redemption
Option (Call):
Optional Redemption
Amount:
The trend of the USD 3-month LIBOR for two years is illustrated in Figure 4.1.
Though it has had an upward movement recently, the USD 3-month LIBOR still
remains a low level. The investors still might have good chance to gain profits from
the inverse floater.
45
US0003M Index
2.5
2
1.5
1
0.5
0
2002/10/1
2003/4/1
Source: Bloomberg
2003/10/1
2004/4/1
2004/10/1
4.1.2 Valuation
I. Construction of Yield Curve
To value the contract, first we have to construct a yield curve on the issue date,
corresponding to the maturity of the contract. In this case of callable inverse floater,
we, at least, construct a 3.25-year (due to the term in arrears)3 quarterly yield curve
on Oct. 04, 2004.
The LIBOR with maturity not longer than one year (3 months, 6 months, 9
months, and one year) is accessible from the market. They are simply the quarterly
yield rates with terms within one year. The LIBOR is shown in Table 4.2.
Libor in Arrears: Libor set in arrears and paid in arrears. At the maturity of the contract, year 3, the
available LIBOR in the market will be L(3, 3, 3.25).
46
3m LIBOR
2.03125%
2y swap rate
2.808%
6m LIBOR
2.21%
3y swap rate
3.368%
9m LIBOR
2.365%
4y swap rate
3.659%
1y LIBOR
2.50375%
For the yield rates with terms longer than one year, we have to get the market data of
swap rate, displayed in Table 4.3. Then we use the Cubic Spline function of Matlab
software to estimate the quarter-united swap rate, i.e. swap rate for 1.25, 1.5, 1.75
years, and so on.
Next, we apply the bootstrapping method to swap rates to obtain the quarterly
yield rate with terms longer than one year. The equation of bootstrapping method is as
j
Sj
i =
1+ S j
1
+
=1
i
(1 + y0,i ) (1 + y0, j ) j
(4.1)
Sj is the swap rate for the interest rate swap terminating at time j; y0,i is the yield to
maturity for time i at time zero. is the time fraction of year. Take for instance the
case when j=1.25. The bootstrapping equation will be as follows.
0.25S1.25
0.25S1.25
0.25S1.25
0.25S1.25
1+ 0.25S1.25
+
+
+
+
= 1.
0.25
0.5
0.75
1
(1+ 0.25y0,0.25)
(1+ 0.25y0,0.5)
(1+ 0.25y0,0.75)
(1+ 0.25y0,1) (1+ 0.25y0,1.25)1.25
In this equation, the only unknown variable is the yield to maturity for time 1.25 at
time zero, y0,1.25. We can solve for y0,1.25 with simple algebra manipulation. Proceeding
recursively, we then bootstrap the entire yield to maturity we need. The yields to
maturity are shown in Table 4.4, and the yield curve illustrated in Figure 4.2.
Term (year)
Yield (%)
Term (year)
Yield (%)
Term (year)
Yield (%)
0.25
2.0313
1.25
2.6020
2.25
2.9391
0.5
2.2100
1.5
2.6681
2.5
3.0915
0.75
2.3650
1.75
2.7301
2.75
3.2503
2.5038
2.8142
3.3937
3.25
3.5046
0.038
0.036
0.034
YTM
0.032
0.03
0.028
0.026
0.024
0.022
0.02
0.5
1.5
2
Maturity
2.5
3.5
(4.2)
L(0, i, j) is the forward rate prevailing between time i and time j known at time zero;
y0,j is the yield to maturity for time j at time zero; , again, denotes the time fraction of
48
year.
Iteratively, we obtain the whole sets of forward LIBOR, which will be used for
simulation. The result of initial forward LIBOR is shown in Table 4.5.
Table 4.5 Initial 3-Month Forward LIBOR L(0, i, j)
Time (year)
Time (year)
[0,
0.25]
2.0313
[1,
1.25]
2.9954
[2,
2.25]
3.9391
[0.25, 0.5]
2.3888
[1.25, 1.5]
2.9985
[2.25, 2.5]
4.4662
[0.5, 0.75]
2.6752
[1.5, 1.75]
3.1024
[2.5, 2.75]
4.8416
[0.75,
2.9203
[1.75,
3.4037
[2.75,
3]
4.9743
3.25]
4.8376
1]
2]
[3,
P ( 0, T ) Black ( L ( 0, T
j
i =1
i 1
, Ti ) , K , Ti 1 vT j cap
First we obtain the swap rates (for the strike price K) and the market quotes of caps on
the issue date of the contract, 2004/10/04, as shown in Table 4.6.
49
Maturity (year)
26.28
30.91
31.22
30.16
28.82
Applying the relationship (3.47), we extract the caplet volatilities implied by the
market quotes of caps. Take for example when j=2,
Maturity (year)
26.28
31.40
26.46
26.21
22.72
0.32
0.31
0.3
Volatility
0.29
0.28
0.27
0.26
0.25
0.24
0.23
0.22
1.5
2.5
3
Maturity
3.5
4.5
50
i 1 t
+ c,
Ti 1vT2i 1 caplet =
Ti 1
( a (T
i 1 t ) + d
e
b(Ti 1 t )
+ c dt ,
Fwd. LIBOR
Fwd. LIBOR
Fwd. LIBOR
L(0,0,0.25)
32.50
L(0,1,1.25)
28.22
L(0,2,2.25)
23.69
L(0,0.25,0.5)
31.45
L(0,1.25,1.5)
27.11
L(0,2.25,2.5)
22.52
L(0,0.5,0.75)
30.39
L(0,1.5,1.75)
25.98
L(0,2.5,2.75)
21.33
L(0,0.75,1)
29.31
L(0,1.75,2)
24.84
L(0,2.75,3)
20.13
L(0,3,3.25)
18.91
51
L(T0 , T j , T j +1 )
1 + L(T , T , T
j =i +1
Tn +1
j +1 )
j (T0 ) i , j dt
(T0 ) ,
dL(T0 , Ti , Ti +1 ) = L(T0 , Ti , Ti +1 ) i ( T0 )
L(T0 , T j , T j +1 )
j =i +1
1 + L(T , T , T
j
j +1
j (T0 ) i , j dt
i (T0 ) are the instantaneous volatilities we recover from the market quotes of caps.
Discretizing these dynamics, then we can start simulate the forward LIBOR at each
time step. The time step is set to be equal to the tenor = 0.25 for simplicity. Due to
the assumption of perfect instantaneous correlation = 1, one random number is
enough for each time step of simulation. Otherwise, we have to generate as many
random numbers as the initial forward LIBOR and perform Cholesky decomposition
for each time step of simulation.
The simulation process is as the following matrix. The first column vector is the
initial forward LIBOR; the second column vector is the simulation for the first time
step t = = 0.25 , and the spot LIBOR L ( 0, 0, 0.25 ) ceases in this time step, so
that it is set to be zero. Repeatedly proceeding, we complete the forward LIBOR
simulation for one time.
52
0
L ( 0,0,0.25) 0
L ( 0,0.25,0.5) L ( 0.25,0.25,0.5) 0
L ( 0,0.5,0.75) L ( 0.25,0.5,0.75) L ( 0.5,0.5,0.75)
L ( 0,0.75,1)
L ( 0.25,0.75,1)
L ( 0.5,0.75,1)
0
0
0
0
0
0
0
0
0
0
0
0
0
0
L ( 2.75, 2.75, 3) 0
This is the illustration of forward LIBOR simulation for one time. It is necessary to
simulate as many times as possible. In addition, we have to record all the forward
LIBOR of every simulation in order to calculate the exercise values.
V. Calculation of Cash Flows
After simulating for M times, we calculate the interest for each payment date
according to the payoff condition in the term sheet. For example, the interest payment
at the maturity of the contract, time 3, is
53
E j (T j ) =
k= j
cashflow ( i, Tk +1 )
1 + Li (T j , Tk , Tk +1 )
i = 1, , M
j=n
(4.3)
where cashflow ( i , Tk +1 ) denotes the cash flow at Tk +1 for the simulation path i. The
simulated spot LIBOR at
Tj
for path i,
Li ( T j , T j , T j +1 )
the
independent
variable
is
and
the
then we obtain the estimate of the exercise values at T j , E j (T j ) . The holding values
for each simulation path at T j , H j (T j ) , is one plus interest payment at T j (nominal
principal plus interest, calculated in percentage). Finally comparing E j (T j ) with
incorporated with the callable feature. Discounting all the cash flows back to time
zero and averaging the M paths results, we have the inverse floater price. When M =
10000, we might have the reasonably theoretical price of 98.87% with variance
54
3.41%.
The valuation procedures may be summarized as follows.
I. Constructing yield curve through market quotes of LIBOR and swap rates.
II. Extracting the initial forward LIBOR implied by the market yield curve.
III. Calibrating the model to the market.
IV. Implementing the Least-squares Monte Carlo simulation of forward LIBOR.
Step IV varies due to the different characteristics of each contract. What makes this
procedure various is the calculation of cash flows. In addition, it is essential to
simulate the forward LIBOR under a single forward measure to ensure the consistent
comparison between the exercise values and the holding values.
Issuer:
Specified
Denomination:
Issue Price:
Issue Date:
27 August 2004
Maturity Date:
27 August 2008
55
Interest Amount:
Period
Coupon Rate
Year 1
5.50%
Year 2
Year 3
Year 4
30/360
Interest Payment
Dates:
Early Redemption
The Issuer may redeem the Notes, in whole but not in part, on each Interest
Option (Call):
Early Redemption
If the Notes are called, the Interest Payment Date in respect of which the
Date:
Optional
Redemption
Amount:
Source: UBS Investment Bank
4.2.2 Valuation
The valuation procedure is very similar to that addressed in Section 4.1.2.
First, we have to construct a 4.5-year semiannual yield curve of LIBOR (Table
4.12). Second, the initial forward LIBOR are extracted from these yield rates (Table
4.13). Using market quotes of caps (Table 4.14) to calculate the volatilities of caplets
(Table 4.15, Figure 4.4), then we are able to recover the instantaneous volatility
structure of forward LIBOR (a = -0.083555, b = -0.035927, c = 0.39781, d =
-0.0032122, Table 4.16).
After all these prerequisites are completed, we start to simulate the forward
LIBOR under forward measure Q 4 . It is necessary to keep tracks of all the simulated
forward LIBOR. The next step is to calculate the cash flows according to the interest
56
payment condition in the term sheet. Discounting the cash flows, running the
regression of the discounted cash flows on the simulated spot LIBOR to estimate of
the exercise values, making the call strategy, changing the cash flow structure, again
discounting the cash flow structure back to time zero, we eventually obtain the
reasonably theoretical price of callable cumulative inverse floater. When M = 10000,
we might have the price of 102.43% with variance 0.24%.
The simulation result shows that this callable cumulative inverse floater is issued
at premium. Observing the simulated forward LIBOR and the cash flow structure, we
find that the interest rate payments are relatively high, due to the low level of
simulated forward LIBOR. Besides, the term of interest payment does not impose any
constraints on the possible highest payment, which might be the key factor causing
the issuance at premium.
Table 4.10 LIBOR on 2004/08/27
6m LIBOR
1.99%
2y swap rate
2.83%
1y LIBOR
2.3%
3y swap rate
3.241%
4y swap rate
3.566%
5y swap rate
3.838%
Term (year)
Yield (%)
Term (year)
Yield (%)
0.5
1.99
2.5
3.064
2.3
3.2624
1.5
2.5845
3.5
3.4404
2.8395
3.6021
4.5
3.7513
57
Time (year)
Time (year)
[0, 0.5]
1.99
[2, 2.5]
3.9645
[0.5, 1]
2.6105
[2.5, 3]
4.257
[1, 1.5]
3.1546
[3, 3.5]
4.5118
[1.5, 2]
3.6065
[3.5, 4]
4.7381
[4, 4,5]
4.9488
Maturity (year)
31.84
34.88
33
30.745
28.76
Maturity (year)
31.84
33.43
27.619
23.579
20.733
0.34
0.32
Volatility (%)
0.3
0.28
0.26
0.24
0.22
0.2
1.5
2.5
3
3.5
Maturity (year)
4.5
58
Fwd. LIBOR
Fwd. LIBOR
L(0,0.5,1)
35.20
L(0,2,2.5)
16.58
L(0,1,1.5)
30.79
L(0,2.5,3)
11.50
L(0,1.5,2)
26.21
L(0,3,3.5)
6.25
L(0,2,02.5)
21.48
L(0,3.5,4)
00.82
L(0,4,4.5)
00.13
Issuer:
UBS AG
Specified
Denomination:
Issue Price:
100.00%
Issue Date:
15 July 2004
Maturity Date:
15 July 2009
59
Interest Amount:
Range
CouponRate1
CouponRate2
Year 1
0 4.00%
1.25%
Year 2
0 5.00%
1.25%
Year 3
0 6.00%
1.25%
Year 4
0 7.00%
1.25%
Year 5
0 7.00%
1.25%
Daycount:
30/360
Interest Payment
Dates:
Early Redemption
Option (Call):
Early Redemption
Date:
Optional
Redemption
Amount:
Source: UBS Investment Bank
4.3.2 Valuation
As mentioned in Section 4.1.2, we start our valuation procedures from building the
daily 3-month yield curve (Table 4.20), extracting the initial daily 3-month forward
LIBOR (Table 4.21), to calibrating the model to the market (Table 4.23, Figure 4.5, a
= -0.086139, b = -0.037914, c = 0.22377, d = 0.14876.)
The daily 3-month yield rates are cubic-splined out of the quarterly 3-month
yield rates. The initial daily 3-month forward LIBOR are extracted from these
60
estimated daily 3-month yield rates. We then simulate the forward LIBOR under the
forward measure Q 5 .
The time step t is set to be 1 360 because of the term of daycount. We have
to know about the simulated forward LIBOR of every day to calculate the interest
payment.
When simulating, we use an indicator variable to count the days when the
simulated spot LIBOR are in the range, and this indicator variable could be used to
calculate the interest payment. However, it needs not to record all the daily forward
LIBOR for estimating the exercise values. What we need is the forward LIBOR on
the interest payment date. For example, the forward LIBOR structure of one time
simulation could be as follows.
L( 0,0,90)
L( 0,1,91)
L(1,1,91)
L( 0,90,180)
L(1,90,180)
L( 90,90,180)
L( 0,180,270) L(1,180,270)
L( 90,180,270)
L( 0,1530,1620) L(1,1530,1620)
L( 90,1530,1620)
L( 0,1620,1710) L(1,1620,1710)
L( 90,1620,1710)
L( 0,1710,1800) L(1,1710,1800)
L( 90,1710,1800)
L(1710,1710,1800)
L( 0,1800,1890) L(1,1800,1890)
L( 90,1800,1890)
L(1710,1800,1890)
L(1800,1800,1890)
We only have to record the forward LIBOR on the 90th, 180th, 270th day, and so on,
with the time interval of 90 days until the maturity of the contract.
Repeatedly estimating the exercise value and comparing it with the holding value
when the contract is callable, we could obtain the reasonably theoretical price. When
61
3m LIBOR
1.62%
2y swap rate
3.021%
6m LIBOR
1.88%
3y swap rate
3.504%
9m LIBOR
2.09%
4y swap rate
3.872%
1y LIBOR
2.31%
5y swap rate
4.153%
6y swap rate
4.609%
Term (year)
Yield (%)
Term (year)
Yield (%)
0.25
1.62
2.75
3.4216
0.5
1.88
3.5331
0.75
2.0913
3.25
3.6397
2.31
3.5
3.7405
1.25
2.5216
3.75
3.8341
1.5
2.7103
3.9191
1.75
2.8809
4.25
3.9952
3.0352
4.5
4.0668
2.25
3.1752
4.75
4.1395
2.5
3.3031
4.2193
5.25
4.3121
Time (year)
Time (year)
[0,
0.25]
1.62
[2.5, 2.75]
3.4216
[0.25, 0.5]
1.88
[2.75,
3]
3.5331
[0.5, 0.75]
2.0913
[3,
3.25]
3.6397
[0.75,
2.31
[3.25, 3.5]
3.7405
1]
62
[1,
1.25]
2.5216
[3.5, 3.75]
3.8341
[1.25, 1.5]
2.7103
[3.75,
4]
3.9191
[1.5, 1.75]
2.8809
[4,
4.25]
3.9952
[1.75,
2]
3.0352
[4.25, 4.5]
4.0668
2.25]
3.1752
[4.5, 4.75]
4.1395
[2.25, 2.5]
3.3031
[4.75,
5]
4.2193
5.25]
4.3121
[2,
[5,
Maturity (year)
25.47
32.66
30.89
28.53
27.07
Maturity (year)
25.47
32.838
25.478
21.035
20.376
0.34
0.32
Volatility (%)
0.3
0.28
0.26
0.24
0.22
0.2
1.5
2.5
3
3.5
Maturity (year)
4.5
63