5
$\begingroup$

In his treatise "An Introduction to Lévy and Feller Processes" (arXiv link), Prof. Dr. René Schilling gives a short and seemingly straightforward proof for the claim that a continuous Lévy process is integrable (Lemma 8.2 on p. 50). More precisely, the claim goes as follows.

Lemma Let $(X_t)_{t\geq0}$ be a Lévy process with càdlàg paths such that $|\Delta X_t(\omega)|:=|X_t-\lim_{s\uparrow t} X_s|\leq c$ for all $t\geq 0$ and some constant $c>0$. Then $E(|X_t|^p)<\infty$ for all $p\geq 0$.

The proof begins as follows.

Proof Let $\mathcal{F}^X_t:=\sigma(X_s,s\leq t)$ and define the stopping times $$ \tau_0:=0,\hspace{1cm} \tau_n:=\inf\left\{t>\tau_{n-1} : |X_t-X_{\tau_{n-1}}|\geq c\right\}. $$ Since $X$ has càdlàg paths, $\tau_0<\tau_1<\tau_2<\cdots$. Moreover, by the strong Markov property, $\tau_n-\tau_{n-1}\sim\tau_1$ and $\tau_n-\tau_{n-1}\perp\mathcal{F}^X_{\tau_{n-1}}$, i.e. $(\tau_n-\tau_{n-1})_{n\in\mathbb{N}}$ is an iid sequence.

The strong Markov property referred to in the proof is the following theorem (Theorem 4.12 on p. 29):

Theorem Let $X$ be a Lévy process on $\mathbb{R}^d$ and set $Y:=(X_{t+\tau}-X_\tau)_{t\geq0}$ for some a.s. finite stopping time $\tau$. Then $Y$ is again a Lévy process satisfying

a) $Y\perp(X_r)_{r\leq\tau}$, i.e. $\mathcal{F}^Y_\infty\perp\mathcal{F}^X_{\tau^+}:=\left\{F\in\mathcal{F}^X_\infty : F\cap\{\tau<t\}\in\mathcal{F}^X_t, \forall t\geq 0\right\}$.

b) $Y\sim X$, i.e. $X$ and $Y$ have the same finite dimensional distributions.

So the strong Markov property can only be applied in the lemma if $\tau_n$ is an a.s. finite stopping time, but why should this be the case?

So suppose, to avoid this caveat, we adjust the underlying probability space and consider only those $\omega$'s for which $\bigwedge_{k=1}^\infty \tau_k(\omega)\neq\infty$. Why should the correspondingly adjusted process $X$ remain Lévy, as required in order to apply the strong Markov property?

$\endgroup$
2
$\begingroup$

If, for all $T$, $P(|X_T-X_0|\leq 2c)=1$, then there is nothing to prove, the process is bounded so it is integrable.

If for some $T$ the above probability $\rho$ is less than $1$, then \begin{align}P(\tau_1=\infty)\leq P(\tau_1\geq NT)&\leq P(|X_T-X_0|\leq2c,|X_{2T}-X_T|\leq 2c,\cdots,|X_{NT}-X_{(N-1)T}|\leq 2c) \\&=\prod_{i=1}^N P(|X_{iT}-X_{(i-1)T}|\leq 2c) \\&=\rho^N,\end{align} by the property of i.i.d. increments, and since $N$ is arbitrary, $P(\tau_1=\infty)=0$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.