Affine forward variance models (Q1999593)

In this paper, the authors propose a class of affine forward variance models which include the Heston stochastic volatility model and the rough Heston stochastic volatility model. They characterise an affine forward variance model by the affine form of its cumulant generating function, which is, in turns, represented as a solution of the convolution Riccati equation. The authors then propose a class of affine forward order flow intensity models which are governed by jump processes and encompass the Hawkes models. They show that the cumulant generating function of an affine forward order flow intensity model is expressed as a solution of a generalized convolution Riccati equation. Section 2 of the paper firstly presents a forward variance model. Then the characterisation of the forward variance model having an affine cumulant generating function is provided in Theorem 2.6, where the affine cumulant generation function is given by the unique global continuous solution of the convolution Riccati equation. It is also noted in Remark 2.7 that the affine cumulant generating function is related to the unique global continuous solution of a nonlinear Volterra equation. The two special cases, namely the Heston model and the rough Heston model, are discussed in the example. The proof of the characterisation result using the martingale property, the Itô's formula, the drift computation and the stochastic Fubini theorem is discussed. Section 3 considers the affine forward order flow intensity models. It is supposed that the probability laws of the random jumps in the models are determined by predictable random measures, where the jump intensity process has a finite expectation and the conditional distribution for the jump sizes satisfies a certain integrability condition. It is also assumed that the order flow processes are self-exciting and have an asymmetric effect on the jump intensity process. Then the characterisation result for the affine cumulant generating function of the affine forward order flow intensity model as the unique global solution of the generalized convolution Riccati equation is provided in Theorem 3.1. Section 4 discusses the high frequency limit of a sequence of affine forward order flow intensity models to an affine forward variance model. The main convergence result based on finite-dimensional marginal distributions is provided in Theorem 4.10.