Stochastic Jacobian and Riccati ODE in affine term structure models (Q2477604)

Affine term structure models (ATSM) with heteroskedastic volatility appear to be at the boundary between traditional linear models and non-linear complex dynamical systems which are considered more realistic for economic modeling of financial markets. These economic motivations find a natural formulation within the framework and the results by \textit{R. J. Elliott} and \textit{J. van der Hoek} [Finance Stoch. 5, No.~4, 511--525 (2001; Zbl 0987.60067)] who show that it is possible to provide a direct relationship between the geometry of the stochastic flows and their Jacobians and the determination of the (deterministic) factor sensitivities, a natural bridge between empirical testing of bond yields and theoretical modeling of macrofactors driving the economy. The proposed approach leads to a deterministic integro-differential equation and does not involve the solution to Riccati equations obtained by the standard Feynman-Kac argument. The authors of this paper find a procedure that reduces such integro-differential equation to a non linear matrix ordinary differential equations. They prove that its solution does necessarily require the solution of the vector Riccati ordinary differential equations. They also provide the link between the stochastic Jacobian approach and the Riccati equations and illustrate the relation between the stochastic Jacobian flow and the linearization procedure for the Riccati ordinary differential equation in the one-dimensional case.