AFFINE PROCESSES 3
Markov process with state space
D
is inﬁnitely decomposable if and only if it is aregular aﬃne process.We also show that a regular aﬃne process
X
is (up to its lifetime) a semimartingale with respect to every
P
x
, a crucial property in most ﬁnancial applicationsbecause the standard model ([53]) of the ﬁnancial gain generated by trading a security is a stochastic integral with respect to the underlying price process. We providea onetoone relationship between the coeﬃcients of the characteristic function of a conservative regular aﬃne process
X
and (up to a version) its semimartingalecharacteristics ([57]) (
B,C,ν
) (after ﬁxing a truncation of jumps), of which
B
isthe predictable component of the canonical decomposition of
X
,
C
is the “sharpbrackets” process, and
ν
is the compensator of the random jump measure. Theresults justify, and clarify the precise limits of, the common practice in the ﬁnanceliterature of specifying an aﬃne process in terms of its semimartingale characteristics. In particular, as we show, for any conservative regular aﬃne process
X
= (
Y,Z
) in
R
m
+
×
R
n
, the sharpbrackets and jump characteristics of
X
dependonly on the CBI component
Y
. This completes and extends the discussion in [31],where they provide suﬃcient conditions for aﬃne diﬀusion (and hence continuous)Markov processes to be well deﬁned and classify them by the number
m
of
Y
i
sthat can enter the conditional variance matrix. We also provide conditions for theexistence of (partial) higher order and exponential moments of
X
t
. An extensionof the main results for the timeinhomogeneous case is given in [47].Some common ﬁnancial applications of the properties of a regular aﬃne process
X
include:
•
The term structure of interest rates.
A typical model of the price processes of bonds of various maturities begins with a discountrate process
{
L
(
X
t
) :
t
≥
0
}
deﬁned by an aﬃne map
x
→
L
(
x
) on
D
into
R
. In Section 11, we examineconditions under which the discount factor
E
e
−
ts
L
(
X
u
)
du

X
s
is well deﬁned, and is of the anticipated exponentialaﬃne form in
X
s
. Someﬁnancial applications and pointers to the large theoretical and empirical literatures on aﬃne interestrate models are provided in Section 13.
•
The pricing of options.
A put option, for example, gives its owner the rightto sell a ﬁnancial security at a prearranged exercise price at some future time
t
. Without going into details that are discussed in Section 13, the ability tocalculate the market price of the option is roughly equivalent to the ability tocalculate the probability that the option is exercised. In many applications,the underlying security price is aﬃne with respect to the state variable
X
t
,possibly after a change of variables. Thus, the exercise probability can becalculated by inverting the characteristic function of the transition distribution
p
t
(
x,
·
) of
X
. One can capture realistic empirical features such as jumps inprice and stochastic return volatility, possibly of a highdimensional type, byincorporating these features into the parameters of the aﬃne process.
•
Credit risk.
A recent spate of work, summarized in Section 13, on pricing andmeasuring default risk exploits the properties of a doublystochastic countingprocess
N
driven by an aﬃne process
X
. The stochastic intensity of
N
([16])is assumed to be of the form
{
Λ(
X
t
−
) :
t
≥
0
}
, for some aﬃne
x
→
Λ(
x
). Thetime of default of a ﬁnancial counterparty, such as a borrower or option writer,