Duplicating and pricing contingent claims with constrained portfolios (Q1593082)

This paper studies the pricing of contingent claims in a market with elastic constraints on portfolio strategies. The underlying \(d\) assets are given by a standard multidimensional Itô process driven by \(d\) Brownian motions; hence the model would be complete in the absence of frictions. Constraints are modelled by a closed subset \(C\) of \([0,\infty)\times\mathbb{R}^d\) and a nonnegative function \(\Phi\) which vanishes only on \(C\). The equation for a self-financing strategy \(\pi\) with wealth \(Y\) then becomes \[ dY_t = \big( r_t Y_t + b_t {\mathbf 1} - r_t \pi_t {\mathbf 1} - \beta\Phi(Y_t,\pi_t) \big) dt + \pi_t \sigma_t dW_t \] so that there is a penalty term with strength \(\beta\) for deviations from the constraints imposed by \(C\). Assuming that \(\Phi\) is Lipschitz, convex and homogeneous of order 1, the authors first show by backward SDE techniques that each nonnegative square-integrable payoff can be duplicated by such a strategy. They then study the resulting time 0 prices and prove that for \(\beta>0\), these prices are convex, sublinear and typically strictly sublinear. The cases of no constraints or rigid constraints are obtained for \(\beta=0\) and as the limit for \(\beta\to\infty\).