Algorithm 963: Estimation of stochastic covariance models using a continuum of moment conditions

We describe the implementation of a parameter estimation method suitable for models commonly used in quantitative finance. The Continuum-Generalized Method of Moments (CGMM) is a Generalized Method of Moments (GMM) type of methodology that applies a continuum of moment conditions to achieve the efficiency of aMaximum Likelihood method. Instead of the transition density, the more commonly available conditional characteristic function is used for estimation. We test the CGMM and a simpler version, called the CMM, on simulated time series to check the recovery of the parameters. We also applied CMM to two stochastic covariance models, the Wishart Affine Stochastic Correlation (WASC) model and the Principal Components Stochastic Volatility (PCSV) model. This illustrates the power of CGMM, as stochastic covariance models are generally hard to estimate. The estimation method is fully implemented in MATLAB.