Professional Documents
Culture Documents
NGOC-KHANH TRAN
SPRING 2018
Learning objectives
Grading
Communications
Topic coverage
Course Objectives
3
Mortgage-backed securities
Corporate bonds
Credit derivatives
US Fixed-income market: outstanding ($ Billions)
11
Source: https://www.sifma.org/resources/research/us-bond-market-issuance-and-outstanding/
US Fixed-income market: Trading volume
12
Source: https://www.sifma.org/resources/research/us-bond-market-trading-volume/
Bond Basics
13
Coupon payment:
No coupon payment
A single CF: face value F (usually $1,000) paid at
maturity T
Price P quoted as a fraction of face value F
Coupon amounts
Coupon rate = fraction "c" of face value paid annually
Annual coupon value=“C” denotes annual dollar coupon
Compounding periods: k
payments are compounded k times each year
Length of each compounded period is (1/k) year
notes.
Sources: http://zfacts.com/p/318.html
US Treasure Market: Recent
22
Maturity
T-Bills (quoted at a discount): up to 1 year
4 weeks.
3-month bills are issued with 13 weeks or 91 days to
maturity.
6-month bills are issued with 26 weeks or 182 days
to maturity.
T-Notes (semi-annual coupons): 1 to 10 yrs.
T-Bonds (semi-annual coupons): over 10 yrs.
US Treasury Market: Classification
24
securities.
Off-the-run: not the most recently issued bond,
brokers.
Biggest volume of trade (2x retail market) is between
dealers. A typical trade is $10-$100 million in this market.
US Treasury Market: Market Structure
26
Bank of Nova Scotia, New York Agency. J.P. Morgan Securities LLC.
BMO Capital Markets Corp. Merrill Lynch, Pierce, Fenner &
BNP Paribas Securities Corp. Smith Incorporated.
Barclays Capital Inc. Mizuho Securities USA Inc.
Cantor Fitzgerald & Co. Morgan Stanley & Co. LLC.
Citigroup Global Markets Inc. Nomura Securities International,
Inc.
Credit Suisse Securities (USA) LLC
RBC Capital Markets, LLC.
Daiwa Capital Markets America Inc.
RBS Securities Inc.
Deutsche Bank Securities Inc.
Societe Generale, NY Branch.
Goldman, Sachs & Co.
TD Securities (USA) LLC.
HSBC Securities (USA) Inc.
UBS Securities LLC.
Jefferies LLC.
Wells Fargo Securities, LLC.
Sources: https://www.newyorkfed.org/markets/primarydealers
US Treasury Market: Individual Investors
28
Who Sets Interest Rates: The market does, but rates are
influenced by government policy.
The Fed controls several key short-term rates:
Discount rate (rate for loans from Fed to temporarily illiquid
banks)
Federal funds rate target (Inter-bank overnight lending rate)
Fed keeps rate near target by buying/selling Treasury bonds
(injecting/withdrawing reserves): buy bonds to decrease
interest rates
“Quantitative easing” (QE): print money and buy
securities, e.g., Treasury securities (but more recently, also
buy MBSs)
US Treasury Market: characteristics
30
𝐵𝐸𝑌 2 4.67% 2
EAY: y= 1 + −1= 1 + − 1 = 4.73%
2 2
(see Excel file enclosed, posted on Blackboard)
Example: yield conventions for zero coupon bonds
38
1000
YTM: 870 = → 𝑦 = 14.94%
1+𝑦
1000
870 = 𝑦 38
→ 𝑦 = 0.734%
1+ 2
𝐵𝐸𝑌 2 0.734% 2
EAY: y= 1 + −1= 1 + − 1 = 0.736%
2 2
Yields: Excel Functions and online calculators
39
Yield curves
Duration and Convexity
Managing Duration & Convexity Risk
Yield curves
41
(*) “Spot” means the curve always plots the current yield (starting from
now/today) to some future date of maturity (instead of from future to future).
Constructing yield curves
42
𝐹 𝐹 1/2𝑇
P= 𝑦 →𝑦 𝑇 =2× −1
(1+ )2𝑇 𝑃
2
BEY(%) 2.02 5.20 6.39 7.09 7.59 7.80 8.00 8.20 8.25 8.49
Pricing using yield curves
44
Answer:
Let 𝑟1 (0, 𝑇) be the BEY for the period (0,T) (subscript
1 indicates annualized value (i.e., 1 year))
45 45 45 1045
P= + + +
𝑟1 (0,0.5) 𝑟1 (0,1) 2
𝑟 (0,1.5) 3
𝑟 (0,2) 4
1+ 1 + 1+ 1 1+ 1
2 2 2 2
45 45 45 1045
= 2.02% + 5.20% 2
+ 6.39% 3
+ 7.09% 4
= 1037
1+ 2 1+ 2 1+ 2 1+ 2
Risk in FI market: first step
46
Default risk
Credit or default risk (not a concern for US Treasuries)
Liquidity risk
Marketability or liquidity risk (also related to credit risk)
Modified duration
Yield curve risk (1-st step to quantify the risk)
48
n k T
CFi
Price-Yield relation: P( y ) i
i 1 y
1
k
Price P is very sensitive to movement in yield, and yield is volatile (see below)!
Short-term yield Long-term yield
Modified duration
49
Properties:
Modified duration 𝐷𝑚 quantifies how price change with yield
dP PDm dy
2. Macaulay duration
40
𝑖 𝑃𝑉(𝐶𝐹𝑖 )
D= × = 10.87 years
2 𝑃
𝑖=1
9%
3. Modified duration 𝐷𝑚 = 𝐷/ 1 + = 10.40 (years)
2
Duration and change in prices
53
𝒅𝑷 𝒚 𝟏 𝒅𝑷 𝟏 𝟏 𝒅𝟐 𝑷
% change in price: = 𝒅𝒚 + 𝟐 (𝒅𝒚)𝟐
𝑷(𝒚) 𝑷(𝒚) 𝒅𝒚 𝟐 𝑷(𝒚) 𝒅𝒚
Convexity
56
𝑑𝑃 𝑦 1 𝑑𝑃 1 1 𝑑2 𝑃 2
% change in prices: = 𝑑𝑦 + (𝑑𝑦)
𝑃(𝑦) 𝑃(𝑦) 𝑑𝑦 2 𝑃(𝑦) 𝑑𝑦 2
1 d 2P 1 n k T i (i 1) CFi
CV
P dy 2
k2 i2
P i 1 y
1
k
Thus convexity gives second-order correction to price change (see
previous graph)
𝑑𝑃 𝑦 1
= −𝑫𝒎 × 𝑑𝑦 + 𝑪𝑽 × (𝑑𝑦)2
𝑃(𝑦) 2
Convexity and change in prices
57
Approximately:
𝑑𝑃 𝑦 1
= −𝐷𝑀 𝑑𝑦 + 𝐶𝑉 (𝑑𝑦)2
𝑃(𝑦) 2
For convexity 𝐶𝑉 > 0, bond price always increases for any fluctuation in
yield (either dy>0 or dy<0).
So given same duration 𝐷𝑀 , higher convexity is better for long position in
bond.
Bullet vs. Barbell Strategies
65
Portfolio (AB) thus trades off lower yield for higher convexity
(compared to the liability).
Nomenclature:
Bullet: Single bond of medium (moderate) maturity
NGOC-KHANH TRAN
SPRING 2018
3.75+100
99.47 = → 𝑦1 = 8.6% (𝐵𝐸𝑌)
1+(𝑦1 /2)
5.5 5.5+100
102.07 = + 𝑦2 2 → 𝑦2 = 8.8%
1+(𝑦1 /2) (1+ )
2
Bootstrapping at work
6
103.75 0 0 0 x1 99.47
5.5 105 .5 0 0 x 102.07
2
4.38 4.38 104.38 0 x3 99.41
5.06 5.06 5.06 105.06 x4 101.02
Bootstrapping: Generality
8
0 x1 P1
0 x P
2 2
0 0 xn Pn
(“×” denotes a given numerical value)
Yield curve risk: next step
9
E.g. 2-2: Consider three zero-coupon bonds. Using the 2yr zero as the
reference bond, the 2-to-10 yield spread is 150 bps, and the 2-to-30 spread is
200 bps. [Assuming annual compounding].
Steepener: You expect the 2-to-10 spread will increase but not sure of the
future overall direction of the interest rate. What is the strategy that
speculates on this view but can hedge against level risk? [Assuming $100
notional value of portfolio]
Compute change in portfolio’s value under parallel shift of +10 bps.
Compute change in portfolio’s value when 𝑦2 drops by 15 bps and 𝑦10
increases by 5bps.
Yield curve risk: example
14
E.g. 2-3: Consider three zero-coupon bonds of e.g. 2-2. Butterfly: the 2yr
and 30yr zeros will move up while the 10yr zero will move in opposite
direction. Not sure about the overall future direction of the interest rate,
nor about the future movement in the slope of the yield curve.
Describe qualitatively your strategy to speculate on this view while
being able to hedge against level and slope risk.
Answer to e.g. 2-3:
2yr and 30yr bond price would drop, while 10yr bond price would surge, so
the speculating strategy is to short 2yr and 30yr bonds, and long 10yr bond.
How to hedge level risk: under level shift, [10yr] and [2yr & 30yr] bond prices
will move in same directions, so short one and long the other in right $
amount will foster delta hedging against this level risk.
How to hedge slope risk: when yield curve steepens or flattens, 2yr and 30yr
bond prices will move in opposite directions, so short both bonds in right $
amount will foster hedging against this slope risk.
Expected inflation and FI market
17
𝐸 𝜋 ≈ 𝑦𝑁 − 𝑦𝑅
Inferring expected inflation
18
Forward rates
Leading theories of yield curve
Basics of forwards and futures contracts
Forward rates: by example
21
Forward rates are interest rates, prevailing from a future point in time
to another future point, & implied by yield curve (and no arbitrage).
E.g. 2-4: Suppose 6-mo BEY 𝑦1 = 6%, and 1-yr BEY 𝑦2 = 8%.
What is the (implied and annualized) interest rate for the period 6-
mo to 1-yr from now?
How a fund can make sure to lend, say, $100 at that rate in future?
Answer (cont):
6 months from now, fund wants to lend (notionally) $100 at (annualized)
rate 𝑓1,2 = 10.02% for 6 months.
They can do that by:
𝑦1
1. Going short (issue/sell) $100/ 1 + in 6-mo bond now and
2
𝑦1
2. Going long (buy) $100/ 1 + in 1-year bond now
2
(So this is a zero net investment strategy now)
So 6mo from now fund will have to deliver (i.e., just like put aside for lending)
to the buyer of 6-mo bond buyer (face & interest of this 6mo bond):
𝑦1 𝑦1
$100/ 1 + × 1+ = $100
2 2
And, 1-yr from now the fund will receive (from long position in 1-yr bond)
𝑦1 𝑦2 2 𝑓1.2
$100/ 1 + × 1+ = $100 × 1 + = $105.01
2 2 2
Then practically, fund is lending $100 at annualized rate 𝑓1.2 for period 6mo to
1yr (from now).
Forward rates
23
𝟏/𝒏
𝟏+𝒚𝒎+𝒏 𝒎+𝒏
𝒇𝟏 𝟎, 𝒎, 𝒎 + 𝒏 =
𝟏+𝒚𝒎 𝒎
Properties:
Forward rates are known now and thus can be locked in
now by trading FI securities (e.g. 2-2).
Forward rates are not future spot rates(*): forward rates
are only an indicator of future spot rates.
Thus forward rates are informative about the market’s
expectation about future interest rate.
Note: In above formula, y and f are effective per-period rates. (Rates
on bond equivalent basis (BEY) must be divided by 2 first to get to 6-
month period rate, see e.g. 2-2)
(*) Future spot rate: is a rate that will be realized in a future date, and
cannot be known for sure any time before that future date.
Yield curve from forward rates
25
Liquidity preference
Convexity effect
27
100
Pricing: 𝑃0 = 1+𝑦1 × 1+𝑓1 0,1,2
(1)
If market expects that at t=1, 1-yr yield (called future spot yield) is 𝑦1,2, the value of
100
bond at time t=1 is 𝑃1 =
1+𝑦1,2
𝑃 100
Thus at time t=0, 𝑃0 = 1 = (2)
1+𝑦1 1+𝑦1 × 1+𝑦1,2
Comparing Eqs. (1), (2) would imply 𝑓1 0,1,2 = 𝑦1,2, but as can only have some
expectation about future spot rate 𝑦1,2, EH postulates that
𝑓1 0,1,2 = 𝐸0 𝑦1,2
Expectation hypothesis (EH)
31
𝑓𝑛 0, 𝑚, 𝑚 + 𝑛 = 𝐸0 𝑦𝑚,𝑚+𝑛
𝑓𝑛 0, 𝑚, 𝑚 + 𝑛 = 𝐸 𝑦𝑚,𝑚+𝑛 + 𝐿 0, 𝑚, 𝑚 + 𝑛
That is, the forward rate perceived by investors goes beyond
the expected future spot yield because of their liquidity
preference.
Liquidity preference in the data
36
𝒚𝒕 𝟐𝒕
𝑭𝑷 = 𝑷𝟎 × 𝟏 + 1
𝟐
Forward price 𝑭𝑷 can also be conveniently
computed from effective annual yield EAY.
Note that, even the coupon bond has not yet come to live, market
can already start trading on it: this is a great feature of forward
market to facilitate borrowing/lending/speculating in future rate
NOW!
So actually, market trades the ownership of all CFs to be produced
by the 10yr coupon bond.
Thus, the current price 𝐏𝟎 of such (not-yet-born) coupon bond is
simply the present value of all future cash flows of this (not-yet-
born) 10yr coupon bond!
Forward price: example
44
At t=2, pay back the loan plus 2-yr interest: 𝟏𝟎𝟎 × 𝟏 + 𝟒. 𝟐% 𝟐 = 𝟏𝟎𝟖. 𝟔
Thus, at t=2, make a sure profit of 𝟏𝟏𝟎 − 𝟏𝟎𝟖. 𝟔 = $𝟏. 𝟒 from a zero-net
investment at time t=0.
Forward price: example
46
Remarks:
Forward price is NOT the price (cost) of the forward contract: price of
forward contract is actually zero at the signing date: it is a fair price to both
parties looking forward.
As time evolves, price of bond unfolds, and at deliver date, the bond actual
spot price eventually will be higher or lower than the forward price. Thus
one party wins some money, the other loses by exactly that amount of
money. Forward contracts are zero-sum game!
NGOC-KHANH TRAN
SPRING 2018
How much do you sell the bills back for the next
day?
Repo: example
9
Answer to E.g. 3-1: You purchase $1,000,000 face value in T-bills for $975,000
in an overnight repurchase agreement quoted at 8.75%.
Assume a haircut of 2%, what is the fair market value of the amount of T-
bills purchased?
We are lending $975,000 and accepting T-bills of 2% more than this value as
collateral. Thus the market $ amount of T-bill used as collateral (i.e.,
purchased) is
𝟏 + 𝟐% × 𝟗𝟕𝟓, 𝟎𝟎𝟎 = $𝟗𝟗𝟒, 𝟓𝟎𝟎
(Note: we can proceed a bit differently: call X the fair market value of the T-bills used as collaterals,
2% of which will be deducted as the haircut: so 𝟏 − 𝟐% × 𝑿 = $𝟗𝟕𝟓, 𝟎𝟎𝟎 → 𝑿 = $𝟗𝟗𝟒, 𝟗𝟎𝟎)
Repo: example
10
institutions).
Financing dealer inventories (e.g., inventories can be used as
collaterals to borrow money for dealers)
Obtaining cash while avoiding realizing capital gains or
losses from selling securities.
Adjusting the money supply (by the Fed)
Creating "synthetic" forward positions.
Sometime, firms go to Repo market not to borrow money, but
to get securities (e.g., to sell short)
Similarly, Speculating on interest rate and spread movements
(usually by dealers and arbitrageurs).
Using Repos to create synthetic forward contract
12
E.g. 3-2: Use Repo market, create the forward contract that
allows you receive (i.e., long) 3-yr bond in 6 months.
Answer to E.g. 3-2 :
Idea: We want to receive (get back) bonds in 6 mo. So, at t=0
enter the 6-mo repo contract as a money borrower.
Thus, give up (3.5-yr) bonds now, and repurchase (get back)
(3-yr) bonds in 6mo at predetermined (forward) price
(which can be computed from the quoted interest rate and
haircut as in e.g., 3-1)
Using Repos to create synthetic forward contract
13
Futures on Eurodollars
Futures on T-bonds
Duration hedging using Forwards &
Futures
Standard futures contracts
15
Futures on T-bonds
Preliminaries:
Eurodollars: are USD deposits (e.g., CD account) to a
banks outside US. So Eurodollars are USD bonds but
outside Fed’s jurisdiction.
LIBOR (London Interbank Offered Rate): is the average
(surveyed) interest rate that London’s leading banks
would pay for borrowing from other banks.
(“surveying” is subject to possible coordinated manipulation: 2012
Libor scandal)
3-mo Eurodollar Futures
16
Answer: It depends on the path of the 90-day LIBOR rate in the market
LIBOR (i) Daily change ($) Cummu change ($)
Day 1 2.70% -125 -125
Day 2 2.71% -25 -150
Day 3 2.65% 150 0
Day 4 2.68% -75 -75
Day 5 2.63% 125 50
Managing duration using futures
19
E.g. 3-4: A bank has a liability of 100,000 USD face due 6mo
from now. The current 6-mo BEY is 7%. Fund is concerned that
this yield will change by the time the liability due date arrives.
How fund can hedge against duration risk for this liability
using 3-mo Eurodollar futures?
Futures on T-bonds :
Deliver (qualified) T-bonds of face value of 100,000 USD
Delivery months are Mar, Jun, Sep, Dec
but with considerable flexibilities (to the short party of the future
contract):
Short party can choose to deliver any T-bonds with time to maturity (or
to first call) longer than 15 years (as of delivery date), and this choice
can be made at delivery date (Why? Because there is a universe of T-bonds
outstanding, and with time-to-maturity larger than 15 years, they have characteristics similar enough)
Short party can choose when in delivery month to deliver the bond
T-bond Futures: flexibility
23
thus short party will wait till delivery date to choose the
then-cheapest (lowest 𝑷𝒕 ) qualified bond to deliver.
T-bond Futures: conversion factor Cf
24
10−5 accounts for the facts that Cf is bond price per $1 par, while T-
bond Futures are contract on 100,000 USD par (because in above formula, 𝑪𝑭𝒊
are cash flows of bond with 100,000 USD face, which is to be delivered in the futures
contract)
T-bond Futures: settlement
25
𝐬𝐞𝐭𝐭𝐥𝐞𝐦𝐞𝐧𝐭 = 𝑭𝒇 × Cf
−5
𝐶𝐹𝑖 𝐶𝐹𝑖
𝐹𝑓 × Cf−𝑃𝑡 =𝐹𝑓 × 10 × 𝑖
− 𝑖
1.03 𝑦𝑡
𝑖 𝑖 1+
2
T-bond Futures: settlement
27
−5
𝐶𝐹𝑖 𝐶𝐹𝑖
𝐹𝑓 × Cf−𝑃𝑡 =𝐹𝑓 × 10 × 𝑖
− 𝑖
1.03 𝑦𝑡
𝑖 𝑖 1+
2
Still,
as conversion factor Cf is not exactly
proportional to spot bond price 𝑃𝑡 , short party still
has some room to pick up the cheapest bond for
delivery
T-bond Futures: example
28
E.g. 3-5: Suppose that the Treasury bond futures price is $102.25 per $100
face (i.e., $102,250 per $100,000 face). At the delivery date, the short party
of the future contract chooses to “deliver” 9%-coupon T-bond with 24
years to maturity. Also on the delivery date, the spot BEY of this bond
turns out to be 7%
Compute the conversion factor Cf
Upon delivery of the bond, how much money does the short party
receive from the long party?
What is the net payoff the short party earns at delivery date?
Instead, on delivery date, the spot BEY of this bond turns out to be
only 5%. Compute net payoff to short party on delivery date?
T-bond Futures: example
29
The short party needs to deliver the bond to the long party.
Given the info (24-yr 9% -coupon Bond yielding 7% BYE), the
price of this bond is $1.23 per $1 face (or $123,000 per $100,000
face).
Thus, on the delivery date, the net payoff to the short party is
the difference between what it receives and what it has to give
up (to the other counterparty in the future contract), which is:
At this new spot BEY=5%, the 24-yr 9%-coupon bond price is $1.56
per $1 face (or $156,000 per $100,000 face) on delivery date.
Note: Change in spot BEY does not affect either conversion factor, or
the forward price specified in contract, hence the future settlement
remains at $𝟏𝟎𝟐, 𝟐𝟓𝟎 × 𝟏. 𝟑𝟖 = $𝟏𝟒𝟏, 𝟏𝟎𝟓
E.g. 3-5: Consider a floating rate bond with semiannual payment. The current
BEY 𝑦𝑖 = 7%. Compute the modified duration of this bond.
Answer to e.g. 3-5:
At any period i, can treat the floating rate bond effectively just as a zero-
coupon bond, with effective face being 𝑭 + 𝑪𝒊+𝟏 = 𝑭 + 𝑭 × 𝒚𝒊 (why? This
is because 𝑭 accounts for all future coupons after 𝑪𝒊 )
Thus, as for all zero coupon bond, the current (year i) modified duration
of the floating rate bond is
6𝑚𝑜 0.5
𝐷𝑖,𝑚 = = = 0.483 yr
1 + 𝑦𝑖 /2 1 + 3.5%
Note: because spot yield 𝑦𝑖 varies over time, so does the duration of
floating rate bond.
Interest rate swap
36
Pricing interest rate swap (what follows applies at the signing date only):
Fixed-rate payor shorts coupon bond at par (as it pays fixed), longs a floating rate
bond (as it receives floating)
Floating-rate payor shorts floating rate bond (as it pays floating), longs a coupon
bond at par (as it receives fixed)
Thus, at the signing date (recall that price of floating rate bond is always the par, or
notional principal 𝑭 specified in the swap contract)
𝑷𝑭𝒊𝒙𝒆𝒅 = 𝑷𝑭𝒍𝒐𝒂𝒕𝒊𝒏𝒈 = 𝑭 or more specifically,
−𝟏
𝒄×𝑭 𝒄×𝑭 𝒄×𝑭+𝑭 𝟏 𝟏 𝟏 𝟏
+ 𝟐
+ ⋯ + = 𝑭 ⟹ 𝒄 = 𝟏 − + + ⋯ +
𝟏 + 𝒚𝟏 𝟏 + 𝒚𝟐 𝟏 + 𝒚𝑻 𝑻 𝟏 + 𝒚𝑻 𝑻 𝟏 + 𝒚𝟏 𝟏 + 𝒚𝟐 𝟐 𝟏 + 𝒚𝑻 𝑻
Swap duration
40
E.g. 3-6: Consider a new 5-year interest rate swap, offering a fixed rate of
6% (semiannual payments), and a floating rate of 6-month LIBOR.
What is the modified duration from the perspective of the fixed rate
payor?
What is the modified duration from the perspective of the floating
rate payor?
Answer to 3-6:
5-year fixed rate bond with a 6% (semiannual) coupon selling at par
has a modified duration of 4.26 yr (using online calculator).
The modified duration of the floating rate bond is
6𝑚𝑜 0.5
= = 𝟎. 𝟒𝟖𝟓 𝒚𝒓
1+𝑦/2 1+6%/2
Hence, the duration to fixed-rate payor (who longs floating and shorts
fixed) is 𝟎. 𝟒𝟖𝟓 − 𝟒. 𝟐𝟔 = −𝟑. 𝟕𝟕 𝒚𝒓
Hence, the duration to floating-rate payor (who longs fixed and shorts
floating) is 𝟒. 𝟐𝟔 − 𝟎. 𝟒𝟖𝟓 = 𝟑. 𝟕𝟕 𝒚𝒓
Use of swap: Duration/Maturity restructuring
41
E.g. 3-7: It is December 18, 2014, and: Southwest S&L finances (i.e., lends)
$10 mil in new 10-year mortgages.
The current mortgage rate is 10% per year, fixed (i.e., SW (Southwest)
receives this rate in this mortgage loan from mortgage holder).
The current 3-month rate on time deposits is 8%.
If the mortgages are financed with 3-month time deposits (CDs) (i.e.
SW borrows $10 mil from 3-month CDs to lend to the mortgage holder),
a profit of 2% is locked in over the first three months.
But after that, the bank bears the risk that CD interest rates might rise.
How SW can hedge this risk?
Answer to e.g. 3-7: This balance sheet risk can be hedged very effectively
using interest rate swap.
Assets Liabilities
10-yr mortgage: $10 mil Debt: $9.5 mil in 3-mo (CD)
deposits
Equity: $0.5 mil (owes this to SW
shareholders)
Use of swap: Hedging balance sheet risk
43
Answer to e.g. 3-7 (cont.): But this swap is not a perfect hedge
for SW for several reasons:
Mortgages are usually amortized over their lifetime, so that
the principal balance is declining.
The frequency of the mortgage payments (often monthly)
does not match the semiannual frequency of the fixed
payments on a plain vanilla swap.
The three month rate paid to depositors (CD) does not
match the six month LIBOR rate received in the swap.
Mortgages can usually be prepaid (accelerated payment).
This leads to the need/advent of customized (non-standard)
swaps. E.g., Amortizing swaps, Swaptions, etc.
Note: Specialized swaps can be more expensive because they require more
paperwork and a counterparty may be harder to locate.
Use of swap: Customization
46
E.g. 3-8: Firms A and B both seek better financing for 5-year loan
Firm A has an improving credit rating, and prefers to borrow at a
floating rate.
Firm B has a worsening credit rating, and prefers to borrow at a fixed
rate.
If borrowing directly from the market, firms are offered the following
rates at the best
NGOC-KHANH TRAN
SPRING 2018
later on, market will correct this and the spread will become
“normal” (convergence)
Speculating on spread movements: the risk
6
P&G bet (i.e., taking excessive speculative position) in the interest rate
swap market led to $157 mil loss.
Transactions: Two 6-mo swaps, Notional value = $100 mil each. One on
German Bund (German Gov Bond); one on US Dollar
30-yr bond yield drops from 6.5% to 5.5%, new bond price
𝑃𝑛𝑒𝑤 = 𝑃 + 𝑑𝑃 = 𝑃 − 𝑃𝐷𝑀 𝑑𝑦
= 100 + 100 × 12.8 × 0.01 = 112.8
P&G USD swap: but when interest rates rise at all maturities
Say CMT yield surges from 5% to 6%
30-yr bond yield surges from 6.5% to 7.5%, new bond price
𝑃𝑛𝑒𝑤 = 𝑃 + 𝑑𝑃 = 𝑃 − 𝑃𝐷𝑀 𝑑𝑦
= 100 − 100 × 12.8 × 0.01 = 87.2
Above e.g. 4.1. exemplifies the loss of LTCM in bets on the convergence of
on-the-run/off-the-run bonds’ yields.
Why yields did not converge: 1998’s Russian sovereign default might
have caused great concerns for illiquidity (off-the-run bond) and thus
keep on-the-run bonds highly valuable for extended period of time. This
not only prevented the yields’ convergence, but actually made them more
diverge.
LTCM: look-back
22
Call options on bond gives the buyer the right to buy the bond at the pre-
specified strike price, in exchange for the premium (i.e. the cost to buy the
options) paid up front
Put options on bond gives the buyer the right to sell the bond at the
specified strike price, in exchange for the premium paid up front
An American option can be exercised anytime prior to maturity
An European option can only be exercised on the date it matures
Popular IR derivatives:
callable bonds (bond issuer can buy back bond before maturity)
puttable bonds (bond holder can sell back the bond to bond issuer before
maturity)
caps, collars, and floors (simple IR derivatives with references to LIBOR)
swaptions (fixed or floating payors have option to exit the swap early)
prepayable mortgages
Interest rate options: payoff graph
25
Answer to E.g. 4.2: The option’s strike price is $95 and option’s maturity is at
t=1. Note that
All bonds have (conventionally) face value of $100
Interest rates (yields) are forward-looking (that is, rate given at t applies for
period from t to t+1)
Denote A and B respectively the one- and two-period bonds (from t=0 perspective)
100 100
Bond prices are: 𝑃𝐴 0 = 1+𝑟 = 1+5% = 95.24; 𝑃𝐵 0 = 90.53 (as given)
0
At t=1, option allows buying bond B for $95, which implies the following option payoff
in state u and d at t=1
𝑉 1, 𝑢 = max 94.79 − 95,0 = 0; 𝑉 1, 𝑑 = max 95.69 − 95,0 = 0.69
Interest rate options: pricing by replication
28
By no arbitrage, the value of this portfolio must also equal the value of
the option at time t=0. That is, the price of the option at time t=0 is
𝑉 0 = 𝑥𝐴 × 𝑃𝐴 0 + 𝑥𝐵 × 𝑃𝐵 0 = −0.73 × 95.24 + 0.76 × 90.53 = 𝟎. 𝟐𝟎
Interest rate options: risk-neutral pricing
29
E.g. 4.3: (continued with e.g. 4.2) First, compute the risk-neutral
probabilities 𝑃𝑟𝑜𝑏(1, 𝑢) and 𝑃𝑟𝑜𝑏(1, 𝑑) for states at time t=1.
Then compute option price.
Interest rate options: risk-neutral pricing
30
Answer to E.g. 4.3: We find the risk-neutral probabilities based on the two-
period bond price today and next period, and the one-period risk-free rate.
𝑃𝑟𝑜𝑏 1,𝑢 ×𝑃𝐵 1,𝑢 +𝑃𝑟𝑜𝑏 1,𝑑 ×𝑃𝐵 1,𝑑
𝑃𝐵 0 =
1+𝑟(0)
𝑃𝑟𝑜𝑏 1,𝑢 ×94.79+[1−𝑃𝑟𝑜𝑏 1,𝑢 ]×95.69
or 90.53 =
1+5%
which implies 𝑃𝑟𝑜𝑏 1, 𝑢 = 0.7; 𝑃𝑟𝑜𝑏 1, 𝑑 = 0.3;
Then the price of option is the discounted value of its expected value
(using the above risk-neutral probabilities)
𝑃𝑟𝑜𝑏 1,𝑢 ×𝑉 1,𝑢 +𝑃𝑟𝑜𝑏 1,𝑑 ×𝑉 1,𝑑 0.70×0+0.3×0.69
𝑉 0 = = = 𝟎. 𝟐𝟎
1+𝑟(0) 1+5%
(These are risk-neutral probabilities because we use risk-free rate 𝑟(0) as discount rate! )
IR option: the non-applicability of Black-Scholes
31
Puttable bonds
Puttable bonds are bonds for which the bond buyer has the right
to sell back (i.e. put) to the bond issuer at pre-specified (strike)
price upon (or after) some pre-specified (exercise) date.
So, when the exercise date comes, bond buyer will sell back (put) the
bond to bond issuer if its strike price is higher than its market value.
Thus, bond buyer holds the put option on the bond.
This feature is an disadvantage to the bond seller, thus it must be
reflected in higher puttable bond price 𝑃𝐵,𝑝𝑢𝑡𝑡𝑎𝑏𝑙𝑒 compared to price
𝑃𝐵 of a plain bond of identical cash flows
𝑷𝑩,𝒑𝒖𝒕𝒕𝒂𝒃𝒍𝒆 = 𝑷𝑩 + 𝑽𝒑𝒖𝒕 𝒐𝒑𝒕𝒊𝒐𝒏 ≥ 𝑷𝑩
Hence, puttable bond must offer lower yield to compensate bond
seller 𝒚𝑩,𝒑𝒖𝒕𝒕𝒂𝒃𝒍𝒆 ≤ 𝒚𝑩
Pricing Callable and Puttable Bonds
35
Pricing strategy
Use interest rate model to price the plain bond (of identical CFs).
Use same interest rate model to price embedded call (or put) option.
E.g. 4.5: Assume the following annual binomial tree for one-period interest
rate evolution with: 𝑃𝑟𝑜𝑏 𝑢 = 𝑃𝑟𝑜𝑏 𝑑 = 0.5 at every node.
Compute the price of a plain 3yr 5.25%-coupon bond (annual coupon
payment)
Assume that this 3yr 5.25%-coupon bond is callable at the end of two
years for $99.50. What is the value of the call option? What is the value of
the callable bond?
Pricing Callable and Puttable Bonds
36
Denote option price V at each node. We find option price is $0.38 at t=0.
E.g. 4.6: All assumptions are as in e.g. 4.5, except now that the
bond can be called in years 1 and 2 (but not in year 0) at $99.50
(thus the call option is of American style). What is the value of
the (American) call option? What is the value of the callable
bond?
Thus option value (at t=0) is $𝟎. 𝟗𝟒, and value of callable bond is 102.08 − 0.94 =
𝟏𝟎𝟏. 𝟏𝟒
Caps, Floors and Collars
40
An interest rate cap is a derivative that (1) specified a cap rate and (ii)
allows the holder (i.e., buyer) to receive payment at the end of each period
in which a floating interest rate (e.g., LIBOR or Treasury yield) exceeds the
cap (or, strike) rate. Cap’s payoff at the end of period t is
An interest rate floor is a derivative that (1) specified a floor rate and (ii)
allows the holder (i.e., buyer) to receive payment at the end of each period
in which a floating interest rate (e.g., LIBOR or Treasury yield) falls below
the floor (or, strike) rate. Floor’s payoff at the end of period t is
An interest rate collar a derivative that specify both cap and floor rates.
Collar allows the holder (i.e., buyer) to receive payment at the end of each
period in which a floating interest rate (e.g., LIBOR or Treasury yield) falls
below the floor rate or exceeds above the cap rate. Collar’s payoff at the end
of period t is
𝐶𝐹𝑡 = 𝑃𝑟𝑖𝑛𝑐𝑖𝑝𝑎𝑙 × 𝑃𝑒𝑟𝑖𝑜𝑑 𝐿𝑒𝑛𝑔𝑡ℎ × 𝑚𝑎𝑥 𝑟𝑡 − 𝑟𝑐𝑎𝑝 , 0 + 𝑚𝑎𝑥 𝑟𝑓𝑙𝑜𝑜𝑟 − 𝑟𝑡 , 0
So collar is simply a portfolio of a cap and a floor
Valuing an interest rate cap
43
E.g. 4.7: Consider a interest rate cap contract associated with a 3-year
floating rate loan for $100,000 indexed to the 1-yr LIBOR. The cap has annual
reset frequency and cap rate is 4.75%. Assume 1-yr LIBOR follows the
binomial evolution depicted below (same as in e.g. 4.5), and
Prob(u)=Prob(d)=0.5 at every node. What is the price of this cap at t=0.
Valuing an interest rate cap
44
First caplet: at node (1,u), cap will payoff because 𝑟𝑢 = 4.98% > 𝑟𝑐𝑎𝑝 =
4.75%. The payoff is 100𝐾 × 4.98% − 4.75% at end of year 2, so after
discounting it back to end of year 1 (node (1,u)), its value is
100𝐾 × (4.98% − 4.75%)
𝑉𝑢 = = 219.1
1 + 4.98%
Second caplet: Similarly, at node uu, and ud this cap will also pay off because interest rates at
these nodes are above 𝑟𝑐𝑎𝑝 = 4.75%.
100𝐾 × (6.76% − 4.75%) 100𝐾 × (5.53% − 4.75%)
𝑉𝑢𝑢 = = 1882.7; 𝑉𝑢𝑑 = = 739.1
1 + 6.76% 1 + 5.53%
The (expected and discounted) value of this second caplet at t=0 is
𝑉𝑐𝑎𝑝𝑙𝑒𝑡2
0.25 × 𝑉𝑢𝑢 0.25 × 𝑉𝑢𝑑 0.25 × 𝑉𝑢𝑑
= + +
(1 + 3.50%)(1 + 4.98%) (1 + 3.50%)(1 + 4.98%) (1 + 3.50%)(1 + 4.07%)
0.25 × 𝑉𝑑𝑑
+ = 433.2 + 170.1 + 171.5 + 0 = 774.8
(1 + 3.50%)(1 + 4.07%)
Thus, the total value of this cap contract at t=0 is
𝑉𝑐𝑎𝑝 = 𝑉𝑐𝑎𝑝𝑙𝑒𝑡1 + 𝑉𝑐𝑎𝑝𝑙𝑒𝑡2 = 105.8 + 774.8 = 𝟖𝟖𝟎. 𝟔
Valuing a swaption
46
We go backward in time from t=2 to find first the optimal exercise policy for the
fixed-rate payor, then we price the swaption accordingly. The pricing is similar to
that of American option: but with swaption, when exercise, (from that
moment on) payoff is actually zero! (with American option: exercise early
because it generates best profits in expectation)
NGOC-KHANH TRAN
SPRING 2018
Source: http://finra-markets.morningstar.com/BondCenter/ActiveUSCorpBond.jsp
Corporate debts: Default
12
Interest rate risk (or yield curve risk): values of corporate bonds change
with market yields, in the same way as T-bond prices changes with yields.
This risk can be hedged using Treasuries or Swaps (recall: immunization,
duration (delta) hedging, convexity (gamma) hedging)
Default risk is the number one risk factor (aside from the “standard”
interest rate risk) in corporate bonds
Default risk is the key difference between corporate bonds and
government (T-) bonds. There is (safely assumed) no default in (US) T-
bonds, yet there is non-zero (market-perceived) possibility of default
even for the corporate bonds of highest ratings
Thus default risk is a key driver behind the (price and) yield spread
between corporate bonds and T-bonds.
The bankruptcy of Lehman was the largest default in history ($120.2 billion debt).
Corporate debts: Default
16
Corporate debts: Default vs. credit risk
17
Recovery rates tend to be low when default rates are high. Why?
Corporate bonds: the payoff claim
22
The simplified picture of payoff to corporate debt holder (i.e., firm’s creditor) at
debt’s maturity T (omitting interests for now):
Solvency: If firm’s asset value A is higher than the face value of debt (or more
specifically, the book value of liability) K, debt holder gets K
Default: If firm’s asset value A is lower than the face value of debt K, debt
holder gets only A
A risky zero coupon corporate debt (bond) is equivalent to a portfolio that includes:
a risk-free zero coupon debt
a short position in a put option on firm’s assets, with strike price equal to the
face value of the debt
This put option is key feature in all risky debt pricing models
Corporate bonds: the payoff claim
23
From the perspective of firm’s equity holders (i.e., firm owner), at debt’s
maturity T:
Solvency: If firm’s asset value A is higher than the face value of debt K,
the firm is solvent, equity holders get: A-K
Default: If firm’s asset value A is lower than the face value of debt K,
equity holders get zero (limited liability!)
Thus equity holders’ payoff is 𝐦𝐚𝐱 𝑨 − 𝑲, 𝟎 , or equity holders have a
long position in a call option on firm’s asset, with strike price equal to
the face value of the debt
Corporate bonds: the payoff claim
24
Note that in case of default, firm’s owner (equity holders) still loses all (zero
payoff). But just because of the backing of guarantor, firm is able to raise debt
(that it was not able to otherwise)
In case of default, guarantor has to make up (ie., pay) for any debt loss to the
creditor: Debt loss = Debt Face – Firm’s Asset value
(omitting fee), payoff to the guarantor
equals short position of a put option on
firm’s asset: i.e., guarantor writes put
option on firm’s asset.
Risky debts: Loan guarantees
27
With loan guarantees, all the risk will be taken by guarantor, while other
creditors enjoy holding a safe bond
Pricing loan guarantees
28
E.g. 5.1: Today XYZ Co. has a market value of $100 million, and next
year the company’s assets will take on one of two values: $140 million
or $70 million. XYZ also has a guaranteed debt with a face value of
$90 million (covering principal and interest), coming due next year.
One-year T-note is currently traded at $95 per $100 face. What is the
lump-sum fee that XYZ has to pay (now) to have this guarantee?
Answer to E.g. 5.1: solution approach
First we compute the payoff to the loan guarantor
Reduced-form models
Structural models
Pricing models of credit risk
32
(reduced-form) to decrease
exponentially with time (age)
Prob(𝜏 > t)=𝑒 −𝜆𝑡 ,
where 𝜆 ≡intensity of default
(Larger intensity 𝜆 means
higher chance of default).
Typically, 𝜆 is small (few % at most)
Technical note: the corresponding
probability density of default is
𝜌 𝑡 = 𝜆𝑒 −𝜆𝑡 (see graph), meaning the probability that firm defaults within interval from t to t+dt is
𝑡
𝜌 𝑡 𝑑𝑡 = 𝜆𝑒 −𝜆𝑡 𝑑𝑡. Then the probability that firms defaults before t is 0 𝜌 𝑠 𝑑𝑠 =
𝑡
0
𝜆𝑒 −𝜆𝑠 𝑑𝑠=1 − 𝑒 −𝜆𝑡 . Thus indeed the probability of survival beyond time t is
𝑡
1− 0
𝜌 𝑠 𝑑𝑠=𝑒 −𝜆𝑡
A reduced-form model of credit risk (100% loss)
34
E.g. 5-2: Let’s consider a risky debt with intensity of default 𝜆 = 10%. In case
of default, the debt holder will lose 𝐿 = 40% of the face value of the debt.
What is the probability the debt will default within next year?
The current 1-year T-bond is currently priced at $95 per $100 face. What is
the price of this risky debt (per $100 face)?
Answer to e.g. 5-2:
The probability the debt will survive through next year is Prob(𝜏 > 1) =
𝑒 −𝜆 = 𝑒 −10% = 90.5% and the probability the debt will default within
next year is Prob(𝜏 < 1) = 1 − 𝑒 −𝜆 = 1 − 𝑒 −10% = 9.5%
The current 1-year risk free rate can be computed from price of T-bond
100 100
95 = ⟹𝑟= − 1 = 5.3%
1+𝑟 95
The bond’s credit spread is 𝑠 = 𝐿 × 𝜆 = 0.4 × 10% = 4%,
and the yield on this risky bond is y = 𝑠 + 𝑟 = 4% + 5.3% = 9.3%
100
Current price of the risky bond (per $100 face) is = $𝟗𝟏. 𝟓
1+9.3%
Structural (Merton’s) model of credit risk
37
This structural model stems from the payoff claim to firm’s asset given the
prospect of firm’s default.
Recall that debt holder’s position is sum of [long position in non-
defaultable debt plus short position in put option on firm’s asset]
𝑷𝟎 = 𝑲𝒆−𝒓𝑻 − 𝑷𝒖𝒕𝟎
Idea: Merton’s structural model employs Black-Scholes option pricing to
compute put price 𝑃𝑢𝑡0 , then follows risky bond price 𝑃0
Assuming firm’s asset value 𝐴𝑡 follows random log-normal process
1 2
𝑙𝑜𝑔(𝐴𝑡 ) = log(𝐴0 ) + 𝜇 − 𝜎 𝑡 + 𝜎 𝑡 × 𝑁 0,1
2
where 𝑁 0,1 is a standard normal random variable (like in Black-Scholes).
𝜇 = mean growth of asset value; 𝜎 = volatility of asset value
At debt’s maturity date T, if 𝐴 𝑇 < 𝐾 (or equivalently log(𝐴 𝑇 ) < log(𝐾)),
firm defaults (because the put ends up in the money, and as bond’s holder
writes the put, he loses money)
At t=0, given the current asset value 𝐴0 , and parameters 𝜇, 𝜎, risk-free rate
𝑟, we know only the distribution of 𝐴 𝑇 (the asset’s random value 𝐴 𝑇 will be
realized only at T)
Structural (Merton’s) model of credit risk
39
1
𝑙𝑜𝑔(𝐴 𝑇 ) = log(𝐴0 ) + 𝜇 − 𝜎 2 𝑇 + 𝜎 𝑇𝑁 0,1 is normal variable with
2
1
mean m = log(𝐴0 ) + 𝜇 − 𝜎2 𝑇 and stddev= 𝜎 𝑇
2
𝐴0 𝜎2 𝐴0 𝜎2
log 𝐾
+ 𝑟+ 2 𝑇 log 𝐾
+ 𝑟− 2 𝑇
where 𝑑1 = 𝑑2 =
𝜎 𝑇 𝜎 𝑇
1 𝐴0
Credit spread 𝒔 = 𝑦 − 𝑟 = − 𝑙𝑜𝑔 1 − 𝛷 −𝑑2 + 𝑒 𝑟𝑇 𝛷 −𝑑1
𝑇 𝐾
Credit spread 𝒔 increases with firm’s asset volatility 𝜎: why?
larger 𝜎 increases chance of default
larger 𝜎 also make put more expensive, giving bond holder (put writer)
higher risk & higher premium since put will more likely be exercised
𝐾
Credit spread 𝒔 increases leverage ratio : why?
𝐴0
larger leverage decreases distance to default, and thus higher chance of
default
Depending on the parameters, credit spread 𝒔 might increase or decrease
with time to maturity T: why?
Larger T increases both stddev 𝜎 𝑇 and asset value’s mean
1
𝜇−2𝜎2 𝑇−log(𝐾/𝐴0 )
Thus, effect on distance to default (DD) in inconclusive 𝐷𝐷 = 𝜎 𝑇
(but when T is large enough, DD increases with T, less default chance, and spread s decreases
with T).
45 Topic 5-3: CREDIT DERIVATIVES
Answer to e.g. 5-3: Let’s consider the CDS contract from the prospect of
the protection seller
The expected and discounted value of his payment (to the protection
buyer, in case the bond defaults) is
𝐷𝑒𝑓𝑎𝑢𝑙𝑡 𝑃𝑟𝑜𝑏 × (𝐿 × 100) 1 − 𝑒 −𝜆 × (𝐿 × 100)
=
1+𝑟 1+𝑟
The expected and discounted value of the payment to him (from the
protection buyer, in case the bond does not default) is (recall: 𝒔 is the fee
seller receives)
𝑆𝑢𝑟𝑣𝑖𝑣𝑎𝑙 𝑃𝑟𝑜𝑏 × (𝒔 × 100) 𝑒 −𝜆 × (𝒔 × 100)
=
1+𝑟 1+𝑟
Fair-value of spread (or CDS fee) 𝒔 is obtained when protection seller
makes even in the CDs contract:
1 − 𝑒 −𝜆 × (𝐿 × 100) 𝑒 −𝜆 × (𝒔 × 100)
=
1+𝑟 1+𝑟
1−𝑒 −𝜆 ×𝐿
⟹𝒔= ≈ 𝝀 × 𝑳 = 10% × 0.4 = 4%
𝑒 −𝜆
Credit default swaps (CDS): reduced-form pricing
49
CDS spread is just the fee paid by protection buyer in CDS contract.
CDS spread characterizes the bond’s riskiness
Bond spread is the difference between the (risky) reference bond’s yield
and riskless T-bond’s yield
so bond spread also quantifies the riskiness of the reference bond
should be (in principle) equally safe. (Indeed, in case risky bond defaults, CDS
protection will pay the face value of the risky bond: Nothing is lost)
Credit default swaps (CDS): Basis trading
51
Before looking at the data, here is the general basis trading strategy:
Negative basis trading (CDS spread < Bond spread): CDS protection tends to be
cheap, it is good to buy bonds (to receive high yields) and buy cheap insurance
(CDS) on it
But is it truly that good? One for e.g. can bet on the convergence of the basis (back
to zero):
If basis indeed gets back to zero,
meaning bond’s yield drop, bond become
more valuable, this is good (because we are
longing bond)
If bonds defaults, it is fine too (because
CDS protection will fully cover the loss)
But if basis continues to drop (but bond does not default), then bond just loses
more and more value and it is not good for long position (specially if one cannot
keep up with the strategy until convergence, i.e. liquidity constraint-remember
LTCM): many firms lost money in convergence bet after 2008.
Thus, sometime, incentive of basis traders is the default of the reference
firm (bond)
Credit derivatives: the dark side
57
Speculative trading on credit derivatives (i.e. when investors have no direct claims
on reference bonds-remember naked CDS) may have important (self-fulfilling)
consequences on reference firm/bond’s fate.
Buying CDS is like buying fire insurance on your neighbor’s house (which may lead
to vile incentive to burn that house to collect insurance payment)
Even when traders have direct claims on the reference bond (for e.g., negative basis
traders indeed also long the reference bond), their ultimate incentive may be the
default of the firm (bond).
When basis traders are also bond holders, they have voting rights in and may
interfere with debt restructuring deals of the firm under distress, with the
ultimate purpose not to rescue firm but make it default
Credit derivative victims:
Richard Fuld (last and long-time serving CEO of Lehman) claims that credit
derivatives destabilized and led Lehman to the its final collapse in 2008
JP Morgan’s Bruno Iksil (London Whale) lost 6 billions in aggressive credit
derivative trades in 2011/2012