| Abstract: | This article develops a direct filtration-based maximum likelihoodmethodology for estimating the parameters and realizations oflatent affine processes. Filtration is conducted in the transformspace of characteristic functions, using a version of Bayes’rule for recursively updating the joint characteristic functionof latent variables and the data conditional upon past data.An application to daily stock market returns over 1953–1996reveals substantial divergences from estimates based on theEfficient Methods of Moments (EMM) methodology; in particular,more substantial and time-varying jump risk. The implicationsfor pricing stock index options are examined. |
