|
Intensity models
(also called reduced form models) are a form of
default model. Lets start by considering what are known as
mortality
models of default. These are essentially a discrete form of
intensity model. Once we have established them, we will take a limit as
time intervals go to zero—and out will pop intensity models.
Mortality models derive their name from their similarity
to actuarial models of human mortality. Define a
survival function s(t).
It might indicate the probability of a human surviving until age t
or the probability of a bond surviving without default for t years.
For the rest of this article, we shall use it to denote the latter.
The probability of a bond defaulting in year t + 1
is given by
This is an unconditional probability. It reflects the
probability at time 0 (when the bond is issued) of default between time
t and time t + 1. If we want the conditional probability of
default—that is, the probability of default between time t and time
t + 1—we apply Bayes' theorem to obtain
 |
[2] |
A survival function can be constructed from historical
bond default data. Constructed in this manner, the survival function
defines a mortality model of default. Exhibit 1 indicates empirical
survival functions by original credit quality. If the numbers were
smoothed, it could reasonably be used to specify a mortality model for
default.
|
|
 |
|
year |
original credit rating |
|
AAA |
AA |
A |
BBB |
BB |
B |
CCC |
|
1 |
1 |
1 |
1 |
0.9988 |
0.9904 |
0.9840 |
0.9565 |
|
2 |
1 |
1 |
1 |
0.9940 |
0.9741 |
0.9354 |
0.8297 |
|
3 |
1 |
0.9965 |
0.9998 |
0.9886 |
0.9350 |
0.8797 |
0.6900 |
|
4 |
1 |
0.9946 |
0.9991 |
0.9827 |
0.9288 |
0.8215 |
0.6338 |
|
5 |
0.9997 |
0.9946 |
0.9988 |
0.9772 |
0.9088 |
0.7727 |
0.6147 |
|
6 |
0.9997 |
0.9946 |
0.9980 |
0.9715 |
0.9002 |
0.7406 |
0.5585 |
|
7 |
0.9997 |
0.9946 |
0.9975 |
0.9645 |
0.8853 |
0.7175 |
0.5330 |
|
8 |
0.9997 |
0.9946 |
0.9966 |
0.9630 |
0.8813 |
0.7024 |
0.5156 |
|
9 |
0.9997 |
0.9943 |
0.9960 |
0.9625 |
0.8659 |
0.6908 |
0.5156 |
|
10 |
0.9997 |
0.9941 |
0.9960 |
0.9602 |
0.8334 |
0.6849 |
0.4942 |
|
|
Empirical survival functions constructed
from default data for 1971-2000. Source: Standard & Poor's. |
Now, instead of considering a one-year time interval,
let's consider an arbitrary time interval
t.
Generalizing [2], the probability of default between time
t and time t +
t,
conditional on their being no default by time t, is
 |
[3] |
We can express this as an average rate of default by dividing by
the time interval
t:
 |
[4] |
Consider an example. Let
t be
three years. Assume s(5) = 0.8921 and s(8) = 0.8609. Then
the conditional probability of default between years 5 and 8 is
 |
[5] |
This is a probability of default over a three year period.
By [4], we convert it to an average annual rate of
default by dividing by 3. The result is an average rate of .0117 defaults
per year over the three-year period.
To obtain an instantaneous rate of default f(t)
at any time t, we take the limit as
t goes
to 0 in [4]:
 |
[6] |
| [7] |
| [8] |
where
is the first derivative of s with respect to t.
The instantaneous rate of default f(t) is
called the default intensity or, to
borrow a word from insurance, the hazard rate.
Intensity models work by assuming some functional form for f(t)
and then calibrating that to current
interest rate spreads.
f(t) can reflect "real" probabilities to support credit risk
management applications. It can reflect
risk neutral
probabilities to support
financial engineering applications. The survival function is recovered
from f(t) by rearranging [8] and
integrating:
 |
[9] |
From this, probabilities of default can be obtained from [1]
or [2] as appropriate.
Note how f(t) plays a role similar to that
of a continuously compounded interest rate in [9]. Use
of default intensities tends to simplify mathematics, which is one reason
intensity models are popular with
financial engineers.
Altman (1989), Asquith,
Mullins and Wolff (1989)
and Altman and Suggitt (2000)
discuss
mortality models of default. The first published intensity model appears
to be Jarrow and Turnbull (1995).
Subsequent research includes
Duffie and Huang (1996),
Jarrow, Lando and Turnbill (1997)
and Duffie and Singleton (1997a,
1997b).
|
|
 |
|
asset value model
A type of default model.
credit derivative
A derivative instrument designed to transfer credit risk from one
party to another.
credit risk Risk due to
uncertainty in a counterparty's ability to meet its obligations.
default model A type of model that assess the likelihood of default by
an obligor.
portfolio credit risk
Credit risk associated with a portfolio of obligations, typically
of multiple obligors.
pre-settlement risk Credit risk of default on a derivative instrument
prior to final settlement. |
|
|
|
 |
 |
Ads by Contingency Analysis
|
|
|
 |
|
Caouette, et al. (1998)
discuss mortality models. Duffie and Singleton (2003) is the
essential book on intensity models. Written by two pioneers in the
field, it discusses use of the models both in financial engineering and
portfolio credit risk measurement. See also Arvanitis and Gregory
(2001)
|
|
|
|
 |
|
Altman
E. I. (1989). Measuring corporate bond mortality and
performance, Journal of Finance, 44 (4), 909-922.
Altman, Edward I., & Suggitt, Heather J. (2000). Default rates
in the syndicated bank loan market - a mortality analysis,
Journal of Banking and Finance, 24(1-2), 229-253.
Asquith, Paul, David W. Mullins and Eric D. Wolff (1989).
Original issue high yield bonds: aging analyses of defaults,
exchanges, and calls, Journal of Finance, 44 (4), 923-952.
Duffie, Darrell,
and Ming Huang (1996). Swap Rates and Credit Quality,
Journal of Finance, 51 (2), 921-49.
Duffie,
Darrell and Kenneth Singleton (1997a). Modeling term
structures of defaultable bonds, Review of Financial Studies,
12 (4), 687-720.
Duffie,
Darrell and Kenneth Singleton (1997b). An Econometric Model of
the Term Structure of Interest-Rate Swap Yields, Journal of
Finance, 52 (4), 1287-1321.
Jarrow, Robert,
Stuart Turnbull (1995). Pricing derivatives on financial
securities subject to credit risk', Journal of Finance, 50 (1),
pp. 53-86
Jarrow, Robert, David Lando and Stuart Turnbull (1997). A
Markov model for the term structure of credit spreads, Review
of Financial Studies, 10 (2), 481-523. |
|
|
|
 |
 |
|
|
|
|
|