On exponential local martingales associated with strong Markov continuous local martingales (Q841482): Difference between revisions

Let \((X_t)\) be any continuous local martingale on some probability space \((\Omega,{\mathcal F},P)\) with respect to a right continuous filtration \(({\mathcal F}_t)\) which starts at \(0\). The continuous quadratic variation of \(X\) is denoted by \(\langle X\rangle\), and \({\mathcal E}(X):= \exp(X-\langle X\rangle/2)\) denotes the stochastic (or Doléans) exponential of \(X\). \(({\mathcal E}(X),({\mathcal F}_t))\) is a continuous local martingale but, in general, is not a martingale. The well-known conditions due to Novikov and, respectively, Kazamaki are sufficient for \(({\mathcal E}(X),({\mathcal F}_t))\) to be a martingale. The main purpose of the present paper is to obtain conditions which are necessary and(!) sufficient for the stochastic exponential associated with a strong Markov continuous local martingale (SMCLM) \((X_t)\) to be a martingale. By definition, \((X,({\mathcal F}_t))\) on \((\Omega,{\mathcal F},P_x(x\in\mathbb{R}))\) is a SMCLM if the following properties are satisfied: (i) \(P_x(X_0= x)= 1\), \(x\in\mathbb{R}\). (ii) \((X,({\mathcal F}_t))\) is a continuous local martingale with respect to \(P_x\) for every \(x\in\mathbb{R}\). (iii) \((X,({\mathcal F}_t))\) is a homogeneous Markov process having the strong Markov property. Based on the study of certain functionals of the form \[ T_t= \int_{\mathbb{R}} L^Y(t, a)m(da),\qquad t\geq 0, \] where \(Y_t:= W_t+ t\) (\(W\) being a standard Wiener process) and \(L^Y\) denotes the local time of \(Y\) (here, \(m\) is any nonnegative measure on the Borel subsets of \(\mathbb{R}\)), the authors show that any SMCLM has a speed measure \(\widetilde m\). The main result (Theorem 4.8) gives, for each \(x\in\mathbb{R}\), a condition in terms of \(\widetilde m\) which is necessary and sufficient for \(({\mathcal E}(X),({\mathcal F}_t))\) to be a martingale with respect to \(P_x\).