Dynamic risk measures: Time consistency and risk measures from BMO martingales (Q928502)

The paper is devoted to dynamic setting for risk measures and to the property of time consistency that is a specific version of backward Markov property. The main result of this paper is a characterization of time consistency for a dynamic risk measure by a ``cocycle condition'' for the minimal penalty. The importance of this characterization is that it is a simple condition which can be easily checked. It is also a crucial property for the construction of new families of time-consistent dynamic risk measures. The key tools for the proof are the dual representation for conditional risk measures and lattice properties. To the particular case of discrete time, it is proved that a time-consistent dynamic risk measure can be simply viewed as a conditional measure on a larger space. Using cocycle condition, the author presents a new class of time-consistent dynamic risk measures constructed from BMO (bounded mean oscillation) martingales. This new class generalizes the dynamic risk measures coming from backward stochastic differential equations and allows jumps.