The Mathematics of Financial Modelingand Investment Management

(Brent) #1

22-Credit Risk Model Derivs Page 697 Wednesday, February 4, 2004 1:12 PM


Credit Risk Modeling and Credit Default Swaps 697

default is endogenous in the BSM model while it is exogenous in the Jar-
row-Turnbull and Duffie-Singleton models. As we will see, specifying
defaults exogenously greatly simplifies the problem because it ignores
the constraint of defining what causes default and simply looks at the
default event itself. The computations of debt values of different maturi-
ties are independent, unlike in the BSM model that defaults of the later-
maturity debts are contingent on defaults of earlier-maturity debts.

The Poisson Process
The theoretical framework for reduced form models is the Poisson pro-
cess.^23 To see what it is, let us begin by defining a Poisson process that
at time t has a value Nt. The values taken by Nt are an increasing set of
integers 0, 1, 2, ... and the probability of a jump from one integer to the
next occurring over a small time interval dt is given by

Pr [ Ntdt + – Nt = 1 ] = λdt

where λ is known as the intensity parameter in the Poisson process.
Equally, the probability of no event occurring in the same time
interval is simply given by

Pr [ Ntdt + – Nt = 0 ] = 1 – λdt

For the time being we shall assume the intensity parameter to be a fixed
constant. In later discussions and especially when pricing is covered in the
next chapter, we will let it be a function of time or even a stochastic vari-
able (known as a Cox process^24 ). These more complex situations are
beyond the scope of this chapter. It will be seen shortly that the intensity
parameter represents the annualized instantaneous forward default prob-
ability at time t. As dt is small, there is a negligible probability of two
jumps occurring in the same time interval.
The Poisson process can be seen as a counting process (0 or 1) for
some as yet undefined sequence of events. In our case, the relationship
between Poisson processes and reduced form models is that the event
which causes the Poisson process to jump from zero to 1 can be viewed
as being a default.

(^23) A Poisson process is a point process. Point processes were briefly introduced in
Chapter 13.
(^24) David Lando, “On Cox Processes and Credit Risky Securities,” Review of Deriv-
atives Research 2 (1998), pp. 99–120. Cox processes were briefly covered in Chapter
13 of this book.

Free download pdf