Backus&Zin/B40.3176.

33/Spring 1999

Advanced Fixed Income Analytics for Professionals

Leonard N. Stern School of Business
April 21, 1999

What Is This Course About?

Quantitative models play an increasingly important role in the nancial services
industry | in valuation, trading, structuring, and risk management. In this course,
you will learn how such models are built and used, put them to work yourself using
proprietary software, develop insights into their strengths and weaknesses, and hear
from experts about industry best practice.
Our approach to modeling introduces state-of-the-art structure with relatively modest technical requirements. Make no mistake: this is a quantitative course. But the
level of mathematics rarely rises as high as high school calculus. Our approach
includes discrete time and continuous states. In our view, this is both a marked
improvement in accuracy over the binomial models typically used in MBA classes
and a substantial gain in simplicity over the stochastic calculus of high-end nancial
theory.
Prerequisite: B40.3333 Debt Instruments and Markets or the equivalent expertise
from other sources. A good introduction is chapter 7 of the manuscript from Debt
Instruments and Markets; see http://www.stern.nyu.edu/~dbackus/dbtclass.htm.

Meeting Times
We meet once a week for 7 weeks: Wednesdays from 5:30 to 8:20pm, March 10 to
April 28 excluding March 17 (spring break).

Grades
Grades will be based on the best 4 of 6 weekly assignments. Students who make an
honest eort to do 4 or more assignments will get at least a B.

Home Page
Most of what you need for the course will be on the course home page:
http://www.stern.nyu.edu/~dbackus/3176
Text les are pdf format, which you can view and print with Adobe's Acrobat Reader
(available free if you don't have it already).

Class Materials
The essential material for the course will be weekly lecture notes (distributed in
class and posted on the Web) and a collection of readings (distributed in class).
We ordered several books, but suggest that you delay buying any of them until a
pressing need arises (if ever):

 Babbel and Merrill, Valuation of Interest-Sensitive Financial Instruments , So





ciety of Actuaries Mongraph M-FI96-1, 1996; a nice book, comprehensive and

not overly technical.
Chriss, Black-Scholes and Beyond , McGraw Hill, 1997; nice review of option
theory by a Goldman manager who teaches at Courant, very clear logical
structure, not overly technical, not particularly concerned with xed income.
Hull, Options, Futures, and Other Derivative Securities (Third Edition) , PrenticeHall, 1997; a standard reference, popular and comprehensive but, in our view,
not as clear as it might be.
Haug, Complete Guide to Option Pricing Formulas , McGraw-Hill, 1998; useful
reference to option pricing formulas.
Rebonato, Interest-Rate Option Models (Second Edition) , Wiley, 1998; more
mathematical than necessary, but covers a lot of ground if you don't mind a
physicist's idea of intuition.

Oce Hours
David Backus: Wednesday, 4:00-5:30pm, and by appointment, Kaufman Education
Center 11-55, (212) 998-0907 and [email protected].
Stanley Zin: Wednesday, 4:00-5:30pm, and by appointment, Kaufman Education
Center 9-58, (212) 998-0722 and [email protected].
Most Tuesdays we will be at The Apple on Waverly, between Greene and Mercer,
starting about 7:30pm. Call or email to verify.

Operating Procedures
 We will start and end class on time and typically take a short intermission.
 Everyone can and will be asked questions in class.
 Assignments will be handed out each class for the next class. They can be







done in groups of up to four.

The syllabus is a forecast: we will modify it (and the assignments) to suit
student interests. A new version will be posted each week.
We are readily available by email. You'll often get a reply within an hour or
two, and almost always within a business day. For best service, choose one of
us at a time.
Assignments are due at the start of class. When appropriate, answers will be
distributed at the same time. Late assignments will not be accepted without
prior arrangement. If necessary, you can fax them to Backus at (212) 995-4212.
Readings are optional. The lecture notes are intended to be self-sucient. The
readings are intended to point you in the right direction should you decide to
pursue any of the topics in more depth.
If we can't answer a question in class, we will do our best to answer it in the
next class.
Material handed out in class will not be available in the next class. You can
get copies of old handouts from classmates or the home page.

Schedule of Classes (subject to change)

Class 1 (March 10): What's a Model (Vasicek)?
 Uses of models
 Spot rates and their properties
 Fundamental theorem of arbitrage-free pricing
 Vasicek model: solution and properties
 Vasicek model: parameter values (\calibration")
 Application to hedging: Vasicek v. duration
 Model assessment: where are the bodies buried?
 Assignment:
{ calibration of Vasicek model to DM interest rate data
{ calibration to 5-year spot rate
 Reading:
{ Backus, Foresi, and Telmer, \Discrete-time models of bond pricing," Sections 2-4, reading package and home page; summary, in more technical
terms than this course, of a variety of popular models expressed in a
common framework.
{ Derman, \Valuing models and modeling value," Journal of Portfolio
Management (Spring 1996), 106-114, reading package; nice nontechnical overview of the role of models in industry.
{ Harvey, Time Series Models (Second Edition) , MIT Press, 1993; nice
book on time series models for those who want a more advanced treatment
of the methods used in the notes.
Classes 2 & 3 (March 24, April 1): Options and Volatility
 Two approaches to valuation
 The Black-Scholes formula
 The term structure of volatility
 Extended Vasicek
 Options on zeros









Options on eurocurrency futures

Caplets and caps
Swap essentials
Swaptions
Normal and log-normal
Delta hedging
Assignments:
{ volatility smile for options on eurodollar futures
{ disaster strikes NatWest (1997)
 Reading:
{ Brenner and Subrahmanyam, \A simple formula to compute the implied
standard deviation," Financial Analysts Journal , September-October 1988,
80-83, and \A simple approach to option valuation and hedging in the
Black-Scholes model," Financial Analysts Journal , March-April 1994, 2528; approximations for implied volatility, delta, etc, that aid computation
and intuition both.
{ Hull, Options, Futures, and Other Derivative Securities (Third Edition) ,
Prentice-Hall, 1997, chs 10 and 11; a good overview of the basic option
model.
{ Jamshidian, \An exact bond option pricing formula," Journal of Finance
44 (March 1989), 205-209; nice treatment, in continuous time, of option
prices for coupon bonds in a Vasicek setting.
Class 4 (April 1): Binomial Models
 Short rate trees
 Two approaches to valuation: recursive and Due's formula
 Calibration to current spot rates
 Replication and hedging
 Applications
 Common variants: Ho and Lee, Black-Derman-Toy, Stapleton-Subrahmanyam
 Model assessment
 Assignment:

6
calibration of Black-Derman-Toy model
valuation of interest rate caps and implied volatilities
 Reading:
{ Backus, \Introduction to state-contingent claims," in Manuscript: Debt
Instruments and Markets , home page and reading package; notes for Debt
Instruments.
{ Black, Derman, and Toy, \A one-factor model of interest rates and its application to treasury bond options," Financial Analysts Journal , JanuaryFebruary 1990, 33-39, reading package; a classic that introduces a workhorse
model in industry.
{ Hull, Options, Futures, and Other Derivative Securities (Third Edition) ,
Prentice-Hall, 1997, ch 15; a good reference to xed income models, despite excessive eclecticism.
{
{

Class 5 (April 14): Comparison of Vasicek and Binomial Models

 Pricing kernels and risk-neutral probabilities
 Comparison of (extended) Vasicek and binomial models
 Where's ?
 Options on zeros
 Exploiting mistakes in binomial models
 Assignment: case tba
 Reading:
{ Backus, Foresi, and Zin, \Arbitrage opportunities in arbitrage-free models of bond pricing," Journal of Economic and Business Statistics , January 1998, home page and reading package; an assessment, in moderately
technical language, of the role of the mean reversion parameter in option
pricing (mistakes you could make with BDT).
{ Backus, Foresi, and Telmer, \Discrete-time models of bond pricing," Sections 5-6, reading package and home page; the basis, in more technical
laguage, of much of the lecture.
Class 6 (April 21): Legendary Mistakes
 Case tba.
 Industry visitor: models and reality

7
Class 7 (April 28): Hull and White
 Mean reversion in trees
 Comparison to Vasicek
 More complex lattice models
 Assignment tba
 Reading:
{ Hull and White, \Numerical procedures for implementing term structure
models I: Single-factor models," Journal of Derivatives , Fall 1994, reading package; introduction to their popular \trinomial" implementation of
the Vasicek model.
{ Hull and White, \Using Hull-White interest rate trees," Journal of Derivatives , Spring 1996, reading package; continuation with comments and
extensions.
Class 7 (April 28): Bells and Whistles
 Multifactor models
 CMS swaps
 Volatility smiles and their interpretation
 \Jumps" and stochastic volatility
 Implications for options
 Delta-hedging revisited
 Volatility trades
 Reading:
{ Backus, Foresi, and Telmer, \Discrete-time models of bond pricing," Sections 7 and 8; reading package and home page.
{ Backus, Foresi, Li, and Wu, \Accounting for biases in Black-Scholes,"
Sections 1-4, home page and reading package; a moderately technical
approach to volatility smiles leading to a very simple result.

Backus&Zin/B40.3176.33/Spring 1999

Assignment 1
Due Wednesday March 24 at the start of class.

1. Calibration to DM rates. The goal is to choose parameters for the Vasicek

model that reproduce the properties of interest rates implied by DM swaps.
(a) Use the DM rates supplied on the home page (the spreadsheet for Lecture 1) to compute the mean and standard deviation of DM rates for
maturities of 1 month and 2, 3, 5, and 10 years. Plot the means against
maturity.
(b) Suppose the autocorrelation of the 1-month rate is 0.92. Choose the parameters of the Vasicek model to reproduce the mean, standard deviation,
and autocorrelation of the 1-month rate and the mean 10-year rate.
(c) (optional) In (b) we intended for you to regard the rates as continuouslycompounded spot rates, just like those we used in class. In fact they were
swap rates. Discuss how you might adapt your calibration to take this
into account.
2. Vasicek for long rates. The classic approach to xed income modeling starts
with the short rate. We could with equal justication choose parameters to
reproduce the properties of (say) the 5-year rate. Using the US treasury rates
described in class:
(a) Choose the parameters of the model to reproduce the mean, standard
deviation, and autocorrelation of the 5-year rate, and the mean of the
10-year rate.
(b) Compare the parameters to those estimated in class. For both sets of
parameters, plot Var (y n ) versus n. Comment.
(Warning: this is harder than it looks. Give it a try but don't get bogged
down if it doesn't seem to be working out.)

Backus&Zin/B40.3176.33/Spring 1999

Assignment 2
Due Wednesday March 31 at the start of class.
Eurodollar volatility smile. We have focussed on the term structure of volatility:
dierences in volatility for options of dierent maturities. Another dimension of
practical interest is \moneyness": dierences in implied volatility across strike prices
for otherwise similar options.

Your mission is to explore this issue with eurodollar futures options. The following prices were reported by Bloomberg on March 16 for June options on the June
contract:
Strike
94.50
94.63
94.75
94.88
95.00
95.13
95.25
95.38
95.50

Call Price
(na)
(na)
0.2225
0.1125
0.0425
0.0225
0.0125
0.0075
0.0075

Put Price
0.0075
0.0150
0.0225
0.0375
0.0925
(na)
(na)
(na)
(na)

For all options, the futures price is 94.955 and the 3-month discount factor is 0.9876.
(a) Compute implied yield volatility for each of the put options. When you do
this, remember to convert the put on the futures to a call on the yield.
(b) For the put at 95.00, suppose the 3-month yield is 4%, rather than 5%, implying a discount factor of b = 0:9900, rather than b = 0:9876. How much
dierence does this make to your estimated volatility?
(c) Compute implied yield volatility for each of the call options. Remember to
use put-call parity to nd the price of a call on the yield.
(d) Compare the volatilities of at-the-money and out-of-the-money calls. If the
call at 95.5 had the same volatility as the call at 95, what would its price be?
(e) Graph implied volatility against the strike price, using dierent symbols to
denote calls and puts. How would you interpret the evidence?

Backus&Zin/B40.3176.33/Spring 1999

Assignment 3
Due Wednesday April 7 at the start of class.
Disaster at NatWest (adapted from Subrahmanyam and Richardson). As derivatives markets develop, we see periodic signs of strain on rms' ability to value complex instruments and manage the associated risks. NatWest's 1997 loss of $90mm
on xed income options is a case in point. The attached articles summarize the
public information surrounding this debacle.

Your job is to write a 2-4 page essay touching on at least 3 of the following issues:
(a)
(b)
(c)
(d)
(e)

How did NatWest lose this much money?

How was the loss concealed for 3 years?
Who was to blame, and why?
Was fraud involved?
What specic problems related to the valuation and hedging of options are
highlighted by this case?
(f) What lessons would you draw for risk management and control?

Backus&Zin/B40.3176.33/Spring 1999

Assignment 4
Due Wednesday April 7 at the start of class.
Interest rate caps in the BDT model. The Black-Derman-Toy model is an industry
standard for xed income valuation | not the most sophisticated model in use, but
a widely-used benchmark. Your mission is to calibrate the BDT model to market
conditions and use it to value interest rate caps of dierent maturities. Current
market conditions include (this is adapted from the quote sheet we handed out):

Maturity
0.5
1.0
1.5
2.0

Spot Rate
4.989
5.129
5.209
5.294

Cap Rate
(na)
5.22
(na)
5.37

Cap Price
(na)
0.14
(na)
0.55

(a) With a time interval of h = 0:5 years and a volatility term structure of
(10; 12; 13; 14) (percent!), construct a 4-period BDT short rate tree consistent
with the spot rates above. What are the state prices?
(b) Compute Y = 6-month LIBOR for each node in the interest rate tree.
(c) A semi-annual interest rate cap generates cash ows of
(1 + Y=200),1(Y , K )+ =2
each period after the current one for every hundred dollars notional principal.
For K = 5:22, calculate the cash ows generated by a one-year interest rate
cap.
(d) For K = 5:37, calculate the cash ows for a two-year interest rate cap.
(e) (optional) Given the cap prices above, how would you adjust the volatility
parameters?

Backus&Zin/B40.3176.33/Spring 1999

Assignment 5
Due Wednesday April 28 at the start of class.
Cancellable swaps. Consider the version of the Black-Derman-Toy model used
in class (eg, p 4-6 of Lecture 4). Your mission is to compute the value of a 1-year
cancellation option on a 2.5-year swap: the ability to cancel the swap at par (namely,
zero) at any date in the rst year of the swap. You might recall from the lecture
that the discount factors and swap rates on which the model is based are:

Period (j ) Disc Factor Swap Rate

1
0.975365
5.052
2
0.949999
5.194
3
0.924837
5.274
4
0.899541
5.358
5
0.874550
5.426
6
0.849939
5.483
(a) Consider the swap as (say) a long position in a xed rate bond and a short
position in a oating rate note. What are the bond's cash ows? Compute its
value for every node in the tree.
(b) If a oating rate note is worth 100, what is the value of the swap in each node
of the tree?
(c) The swap can be cancelled free at any time during the rst year. What cash
ows are generated by cancelling?
(d) Apply the logic of American options to value the option. What premium
should be charged for the cancellation option?