Differentiability of quadratic BSDEs generated by continuous martingales (Q2428052)

The authors consider the following BSDE \[ Y_t= B -\int_t^TZ_sdM_s+\int_t^Tf(s,Y_s,Z_s)dC_s-\int_t^TdL_s+ \frac{\kappa}{2}\int_t^Td\langle L,L \rangle_s, \] where \(t\in [0,T]\), \(M\) is a \(d\)-dimensional continuous local martingale, the generator \(f\) is a quadratic function of \(Z\), \(C\) is a continuous increasing process related to \(M\), \(L\) is a martingale orthogonal to \(M\) and the terminal conditon \(B\) is a bounded smooth function of the terminal value of a forward process \(X\) which is of the form \[ X_t=x+\int_0^t \sigma(s,X_s,M_s)dM_s+\int_0^tb(s,X_s,M_s)dC_s. \] If the local martingale \(M\) is a strong Markov process, the solution processes \(Y\) and \(Z\) satisfy the Markov property. Then, it is proved that the backward process \(Y\) is differentiable with respect to \(x\). This enables the authors to show that there exists a deterministic function \(u\) such that \(Y_s=u(s,X_s,M_s)\) and \[ Z_s=\partial_2u(s,X_s,M_s)\sigma(s,X_s,M_s)+\partial_3u(s,X_s,M_s). \] This formula is relevant in the context of securing random liabilities arising from exogenous risks, which are optimally hedged by investment in a given financial market with respect to exponential preferences. It is designed to extend a genuinely stochastic representation of the optimal replication in cross hedging insurance derivatives from the classical Black-Scholes model to incomplete markets on general stochastic bases. In this setting, Malliavin's calculus which is required in the Brownian framework, is replaced by new tools based on techniques related to the calculus of quadratic covariations of basis martingales.