An ergodic BSDE approach to forward entropic risk measures: representation and large-maturity behavior (Q1711728)

Continuing the work of \textit{T. Zariphopoulou} and \textit{G. Žitković} [SIAM J. Financ. Math. 1, 266--288 (2010; Zbl 1230.91084)], the authors study maturity-independent risk measures in incomplete market models with multiple stocks and stochastic factors. The risk measures are constructed as negative indifference prices under exponential forward performance criteria where the forward performance processes are deterministic functions of the stochastic factors. The stock price processes are \[ \frac{dS_{i}(t)}{S_{i}(t)}=b_{i}(V(t))dt + \sigma_{i}(V(t))dW(t)~(i=1,\dots, n) \] where \(W\) is a \(d\)-dimensional Brownian motion with \(d \geq n\) and the \(d\)-dimensional process \(V\) satisfies \[ dV(t) = \eta(V(t))dt + \kappa dW(t),~V(0) = v. \] The forward performance processes are of the form \(U(x,t) = -e^{-\gamma x + f(V(t),t)}\) where \(\gamma > 0\) and \(f\) is identified as the solution of an ergodic backward stochastic differential equation (BSDE) accounting for risk -- i.e. stock price volatility -- and trading constraints. Then for a risk position \(\xi\) and \(t > 0\), the \textit{forward entropic risk measure} \(\rho(\xi,t)\) is defined implicitly via the expected forward performance of optimal trading strategies. The authors obtain \(\rho\) as the solution of a BSDE, and solve for it explicitly in a special case with \(d=2\). A numerical experiment contrasts the forward entropic risk measure to other entropic risk measures in the case when \(V\) follows the Vasiček model.