We are trying to implement a model proposed by Jiang (2002) [ http://papers.ssrn.com/sol3/papers.c...ract_id=639768 ]including a Heston-type stoch. vol., random jumps (Poisson intensity, lognormal size) and a CIR-type stoch. interest rate process. The benefits of Jiang's model is that we are able formulate a data-generating process (estimated with GMM using exact moments) as well to attain the preference-related parameters in order to price bond and options. The data-generating or fundamental process will be used to simulate long-term trajctories of an equity index and a benchmark interest rate, while we want to be able to price options and bonds from these parameters.
However, we can't seem to get the moment conditions correct, hence fail to employ the GMM estimation for all models including the stoch. vol. and/or random jumps. That is, we get reasonable estimates for the BSM and the CIR models, respectively. We have tried to locate the errors by means of MC simulation (using a Milstein scheme), but fail to do so as we can't seem to get convergence for the "autocovariance moment". All moments are calculated from the unconditional characteristic function in (log(S_t),r_t) and (log(S_t),log(S_t-tau)), respectively.
As we have been working on this model for a couple of months now, we are, to say the least, becoming a bit frustrated. Any input on how we can retrieve the correct moments from anyone who has some experience from this kind of set-up or any further reference on papers employing this model is greatly appreciated.
Thanks in advance!
However, we can't seem to get the moment conditions correct, hence fail to employ the GMM estimation for all models including the stoch. vol. and/or random jumps. That is, we get reasonable estimates for the BSM and the CIR models, respectively. We have tried to locate the errors by means of MC simulation (using a Milstein scheme), but fail to do so as we can't seem to get convergence for the "autocovariance moment". All moments are calculated from the unconditional characteristic function in (log(S_t),r_t) and (log(S_t),log(S_t-tau)), respectively.
As we have been working on this model for a couple of months now, we are, to say the least, becoming a bit frustrated. Any input on how we can retrieve the correct moments from anyone who has some experience from this kind of set-up or any further reference on papers employing this model is greatly appreciated.
Thanks in advance!
